Stock Markets Analytics Zoomcamp 2026

Homework 2: One Dataframe Statistics

Distribution of scores and reported study time for this homework.

Submissions

109

Median total score

11

Average total score

10

Score distribution

All values are points.

Questions score

Min
1
Median
11.0
Max
11
Q1
9.0
Avg
9.9
Q3
11.0

Learning in public score

Min
-
Median
0.0
Max
3
Q1
0.0
Avg
0.3
Q3
0.0

Total score

Min
1
Median
11.0
Max
14
Q1
11.0
Avg
10.1
Q3
11.0

Time distribution

All values are hours reported by students.

Lectures

Min
1.0
Median
2.0
Max
40.0
Q1
2.0
Avg
4.4
Q3
4.0

Homework

Min
0.0
Median
3.0
Max
16.0
Q1
2.0
Avg
4.0
Q3
5.0

Question breakdown

Correctness and answer distribution per question.

1. What is the total withdrawn IPO value (in $ millions) for the company class with the highest total withdrawal value?

103 / 109 correct (94.5%)

1 200 0 (0.0%)
2 300 0 (0.0%)
3 400 6 (5.5%)
4 500 103 (94.5%)

2. What is the median Sharpe ratio (as of 11 September 2026) for companies that went public before 1 September 2025?

95 / 109 correct (87.2%)

1 -0.04 9 (8.3%)
2 0.04 95 (87.2%)
3 0.1 3 (2.8%)
4 0.2 0 (0.0%)

3. What is the optimal number of months (1 to 12) to hold a newly IPO'd stock in order to maximize the median growth value?

94 / 109 correct (86.2%)

1 1 94 (86.2%)
2 3 6 (5.5%)
3 5 2 (1.8%)
4 7 3 (2.8%)

4. What is the total profit (in $ thousands) you would have earned by investing $1000 every time a stock was oversold (RSI < 30)?

97 / 109 correct (89.0%)

1 65 97 (89.0%)
2 85 3 (2.8%)
3 105 4 (3.7%)
4 125 1 (0.9%)

5. How would you change the strategy if you want to increase profitability?

109 / 109 correct (100.0%)

Answer Count
I would filter IPOs by fundamentals, valuation, sector momentum and liquidity, then enter only after price/volume confirmation instead of buying every IPO. I would tune the holding period and risk controls with walk-forward backtesting, position sizing and stop-loss rules. 1
I would replace indiscriminate IPO buying with a selective, testable strategy. 1
Don't buy at the first-day close — enter after the quiet period or at lock-up expiration, filter out micro-cap deals and weak underwriters, require positive momentum after month one, and use stop-loss rules instead of a fixed horizon. 1
Momentum-Confirmation Filter Don't buy at the IPO. Instead: Screen: wait 3 months after listing. Only keep IPOs whose price is above their IPO-day close (positive 3-month return) — this discards the early losers seen in Q3/Q4. Enter: invest $1,000 at the 3-month price for the surviving "eligible" IPOs only. Track: measure return/value at +6m, +9m, +12m from that entry point (using latest available price if the checkpoint is still in the future). Why this should help: early positive momentum acts as a proxy for genuine demand/fundamentals, filtering out the flops that drag down the median in Q3. It trades "buy every IPO" for "buy only the IPOs the market has already validated," at the cost of missing the very first 3 months of any winner's move. 1
The current RSI<30 strategy is profitable but unfiltered — it buys every oversold signal indiscriminately, with a modest 55.13% win rate and 1.26% average 30-day return. A few concrete changes, based on patterns visible in this homework's own results, would likely improve profitability: Avoid combining the RSI strategy with newly-IPO'd stocks. Q3 showed IPO'd stocks lose value on a median basis the longer they're held — even the best holding period (1 month) still had a median return of only 0.94 (a ~6% loss). Filtering out any ticker within its first 6-12 months of trading before applying the RSI<30 signal would likely raise the strategy's win rate, since much of the "oversold" signal on fresh IPOs may reflect genuine post-IPO price decay rather than a temporary dip worth buying. Add a trend/quality filter alongside RSI. RSI<30 alone doesn't distinguish a temporary pullback in a fundamentally sound stock from a stock in genuine decline. Combining RSI<30 with a longer-term moving average filter (e.g., only buy if the stock is still above its 200-day moving average) would filter out "falling knives" and keep only oversold dips within an otherwise healthy uptrend. Use volatility-adjusted position sizing instead of a flat $1,000. Some tickers triggering RSI<30 are far more volatile than others (as seen in Q2's Sharpe ratio spread). Sizing positions inversely to recent volatility (smaller bets on high-volatility names, larger on stable ones) would reduce the impact of the worst outlier losses while keeping the average return intact. Tighten the RSI threshold back toward 25 instead of 30. The homework's own observation notes that lowering the threshold from 30 to 25 last year produced fewer trades (~1,568 vs. ~5,206) — while I don't have the exact win-rate/return comparison, a stricter threshold generally selects for more genuinely oversold (rather than mildly weak) conditions, which is worth testing directly against this year's data to see if it trades off volume for a higher per-trade edge. 1
Invest only in companies with positive Free Cash Flow (FCF) and revenue growth > 20%. 1
NATR (Normalized Average True Range) measures daily volatility as a percentage of price ($100 \times \text{ATR} / \text{Close}$), making cross-stock comparisons possible. Slowk (Slow Stochastic) tracks where the close sits within its recent range (0–100). Together with RSI, they target high-payoff mean-reversions: NATR ensures big moves, Slowk pinpoints range bottoms, and RSI confirms momentum. 1
Cambiaría la estrategia de inversión en IPO incorporando criterios adicionales antes de comprar, como analizar el crecimiento esperado, el comportamiento inicial de la acción, la volatilidad y otros indicadores financieros. También compararía diferentes períodos de mantenimiento para identificar aquellos que presenten mejores resultados y evitaría invertir automáticamente en todas las IPO. Finalmente, probaría la estrategia con datos históricos mediante backtesting para comprobar si las modificaciones realmente aumentan la rentabilidad y reducen el riesgo. Esta pregunta es abierta y la propia tarea indica que se deben proponer ideas para identificar IPO con rendimientos futuros positivos o construir una estrategia más efectiva. 1
Fundamental Quality Screening. Instead of trading all IPOs, apply a fundamental filter. Only consider companies with positive revenue grwoth, a clear path to profitability, and manageable debt-to-equity ratios. Avoid speculative Blank Check companies, which hav ethe worst post-IPO decay 1
To increase profitability, I would filter IPOs by financial health metrics (like revenue growth and debt) before entering a position, and combine RSI with a trend indicator like MACD to avoid false oversold signals. 1
The questions did not profile the stock and market. Results will likely change if some categorization was included. The mean return compared across months proved that data is positively skewed and can be reason that there exist opportunities. The challenge is with identifying which names are likely to become multibaggers within the expected period. The comparison also did not ask for liquidity filters, and I suspect many of the winners were illiquid and untradable (unrealistic). That aside, I'd start with categorizing the stock and market based of fundamentals and sentiment among others. 1
Invert the strategy by Short Selling: Since the vast majority of IPOs deliver consistently negative average and median returns over their first year, the most straightforward approach is to flip the trade: instead of buying (long), adopt a systematic short-selling strategy (or purchase put options) after the initial first-day hype subsides, profiting directly from the typical post-IPO downward drift. 1
To increase profitability, I would improve the strategy by combining RSI with other indicators instead of buying every time RSI falls below 30. For example, I would wait for RSI < 30 along with a positive trend reversal or bullish moving-average crossover. I would also use stop-loss and take-profit levels to control risk, avoid stocks with very low liquidity, and test different holding periods to identify the most profitable one. Finally, I would back test the improved strategy on historical data before using it. 1
To increase profitability in IPO investing, I would implement a **multi-factor filtering and timing approach** rather than buying all IPOs indiscriminately 1
The data shows buying at the IPO close and holding is a losing strategy: the median growth falls from 0.94 at 1 month to 0.48 at 11 months, and only ~38% of the stocks are above water after 1 month. So I would (1) shorten the horizon drastically - exit within 1-2 months instead of holding a year; (2) stop buying every IPO and filter first: skip offers priced under $5 and very small deals, which concentrate the worst outcomes, and prefer deals led by top-tier underwriters; (3) avoid entering on day one at all - wait for the lock-up expiry and the post-IPO drawdown, then combine with the RSI signal from Q4 and only buy when the stock is already oversold (RSI < 30), which had a 55% win rate over 30 days; (4) consider the opposite side: since the median IPO underperforms, a market-neutral version (short the weakest IPOs, long a benchmark like SPY) could capture the negative drift instead of fighting it. 1
I would improve the strategy by adding more filters before entering a trade. For example, I would include fundamental company information, market conditions, volatility, liquidity, trading volume and sentiment indicators. I would also test machine learning models to identify patterns associated with positive future returns and use proper backtesting with time-based validation to avoid overfitting. The goal would be to create a more selective strategy that focuses on higher-probability opportunities instead of increasing only the number of trades. 1
The biggest improvement is to only buy when the stock is already in an uptrend. Right now, the strategy takes every RSI < 30 signal, but most oversold stocks keep falling. If you only buy when the price is above its 20-day average, the win rate jumps from 55% to 60%, and the average profit increases from 1.26% to 1.85%. You get fewer trades, but they are much better ones. The second change is to only buy at truly extreme oversold levels. Dropping the threshold from RSI < 30 to RSI < 20 increases the average return from 1.26% to 3.07%. The trade-off is that you get many fewer signals, but each one earns much more, and the profit per trade is significantly higher. 1
I would make the RSI threshold more selective, for example by buying only when RSI < 25, and test whether this increases the average return and total profitability while maintaining a reasonable number of trades. 1
To increase profitability, I would use a combination of IPO fundamentals, valuation, market conditions, trading volume, and technical indicators rather than investing in every IPO. I would also apply risk-management techniques such as position sizing and stop-loss rules. A machine-learning model could further be developed to identify IPOs with a higher probability of positive future returns based on historical IPO characteristics and market conditions. 1
exit at RSI 50 1
Sell the stock immediately if the price drops by a certain percentage (e.g., 5% or 10%) after buying, which helps cut losses early. 1
I’d screen IPOs for sound fundamentals, reasonable valuation, and strong demand rather than buying all IPOs. I’d wait for volatility to settle, diversify, and test the strategy with transaction costs and delisted stocks included. 1
adding more techical startegie like MACD or Fibonaci trace to support the RSI.This multi-indicator approach helps filter out false signals and increases the probability of successful trades. 1
N/A 1
better entry filters according to market regime 1
I wouldn’t invest in every IPO blindly. I would choose IPOs with additional features like sector, company fundamentals valuation, market trend, trading volume and momentum indicators. I would also try to enter after the initial IPO period or lock up expiration use risk management rules such as stop-losses, and backtest the strategy using historical data before using 1
Only buy the IPOs a model trained on past deals scores well (using what's known at IPO time), wait out the post-IPO drift before buying, and use stop-loss and take-profit exits instead of holding for a fixed period. 1
Don't buy on day one. 1
Instead of buying all IPOs we'll filter companies which have strong financial health and tier one backing 1
I would try to identify IPOs that are more likely to have positive future returns instead of investing in every IPO. I could use features such as the initial return, volatility, trading volume, and market conditions, and then use a machine learning model to predict future returns. I would also test different holding periods and invest only when the predicted return is positive. 1
i would do financial sentiment analysis on that company, track VIX, its probability of entering an index fund ( joining s&p might offer stock price increases) 1
To increase the profitability of an IPO investing strategy, switch to a tactical short-term holding period (such as 1 to 3 weeks) to capture the initial listing pop, apply fundamental profitability screeners (e.g., positive trailing earnings) to filter out lower-quality offerings, and buy only after the 6-month insider lockup expiration clears. 1
To pivot an IPO strategy into positive expected returns, focus on five quantitative adjustments to bypass retail traps and insider dumping:Fundamental Quality Filter: Target IPOs with Rule of 40 $\ge 40\%$ (Revenue Growth + FCF Margin) and positive 12-month pre-IPO cash flow. Stick strictly to Tier-1 underwriters (Goldman Sachs, Morgan Stanley, JPMorgan).Institutional Demand Signal: Only buy IPOs priced ABOVE their initial S-1 range (e.g., filing range $\$18-\$21$, final offer price $\$23$). Ignore those priced within or below.Wait Out Lockup Expiration: Avoid Day-1 buys. Enter 15–30 days AFTER the 180-day lockup expires to let insider selling pressure exhaust, and verify institutional re-accumulation.Post-Earnings Confirmation: Require the company to beat consensus estimates in its first two public quarters ($Q1/Q2$) and raise full-year guidance before entering on a breakout above its initial high base.Asymmetric Risk Controls: Cut losses hard at $-8\%$ to $-10\%$, but ride top winners using a 50-day moving average trailing stop to capture multi-year power-law returns. 1
Instead of trying to predict which IPO will rise (which is difficult), you filter the universe to eliminate structural flaws (SPACs, illiquid stocks, those with lock-up periods) and optimize asymmetry: by limiting losses, not by maximizing gains. 1
Probably add filters for the RSI strategy instead of a blanket buy of every RSI less than 30. 1
refine the RSI strategy instead of buying solely when RSI is below 30 1
I'd add quality filters (skip SPACs/micro-caps, pick profitable firms with top underwriters), wait 1–3 months for the post-IPO dip instead of buying day-1, avoid high-VIX markets, use tight stop-losses with trailing exits for winners, and train an ML classifier (XGBoost on fundamentals + technicals + macro) to only take trades with >60% predicted win probability. 1
Since most IPO strategies show negative average and median returns, I would improve profitability by filtering for higher-quality IPOs rather than treating all IPOs equally. 1
I would not invest in every IPO automatically. Instead, I would build a filtering or prediction model to identify IPOs that are more likely to produce positive future returns. I would use features such as first-day return, trading volume, price momentum, RSI, volatility, market conditions (S&P 500, VIX, interest rates), industry, and company characteristics. I would then train a classification model where the target is whether the stock has a positive return after a chosen period, such as 30, 90, or 252 trading days. I would also combine the prediction with risk management—for example, investing only when the predicted probability of a positive return is above a threshold, avoiding extremely volatile IPOs, and using position sizing or stop-loss rules. The strategy should be evaluated using time-based train/test splits to avoid using future information and compared against a simple buy-and-hold IPO strategy. 1
I would not buy on the first day, because most new IPOs lose value in the first months. Instead, I would wait until after the lockup period and only buy IPOs with a big deal size and a strong market trend. 1
Most IPO strategies show negative median/mean returns (Q3 medians < 1 for every holding period; Q2 growth_252d median 0.60 vs mean 1.06 — a few outliers mask broad underperformance). To increase profitability I would: (1) filter IPOs by quality signals before entry — profitable revenue scale, reputable underwriters, realistic offer pricing vs. peers, insider/lock-up structure; (2) replace blind buy-and-hold with momentum/technical triggers (e.g. only hold while price > 20-day MA, or Q4-style RSI mean-reversion entries after the initial IPO sell-off); (3) cut losers fast (time-stop / trailing stop around month 1–2, where Q3 medians peak then decay) and let the few winners run; (4) size positions by volatility (Sharpe-style risk adjustment from Q2) instead of equal weight. 1
Q3 shows the unconditional strategy loses at every horizon: buy all 133 IPOs at the first day's close and median growth falls from 0.94 at one month to 0.53 at twelve. Splitting the same stocks by deal size (offer price x shares offered) shows that median is an average over two unlike populations. bucket n median 12m mean 12m share > 1.0 small 44 0.165 0.789 20.5% mid 41 0.298 1.029 19.5% large 44 0.921 1.098 43.2% Deal size is bimodal rather than continuous: the median deal is 151M, so "small" and "mid" are micro-cap listings and "large" spans 1.75B. These are different markets, not a size gradient. The buckets are indistinguishable at one month (0.926 to 0.948) and diverge over the following year, so the effect is not the first-day pop. What this does and does not support: Excluding micro-cap offerings removes the segment where the typical stock loses 84% of its value in a year. This is the clearest available improvement. It does not produce a profitable strategy. Large deals still have a median below 1.0 and are profitable only 43.2% of the time. In small and mid buckets the mean sits far above the median, so what return exists is concentrated in a few names. Any strategy there is a tail bet requiring wide diversification, and the median is the wrong tool to judge it. Untested directions: entering at the offer price rather than the first day's close, which separates allocation from trading; and shorting, which the consistently negative medians suggest but which is limited in practice by borrow cost and availability on recent IPOs. These results rest on 133 stocks from a single year, with 41 to 44 per bucket. The direction is large enough to be worth noting; the magnitudes are not stable estimates. 1
IPO returns are very uneven: the median 2025 IPO lost about 47% over a year, while a handful multiplied several times and dragged the mean back to roughly break-even. So a strategy has to either catch those few winners or avoid the many losers. First, I would use information the price data doesn't contain: sector, size of the offering, whether the company is profitable, whether it is a SPAC, and country of origin. That could feed a model predicting which IPOs end the year up, or just rule out categories that have done badly. Second, I would filter out the problem cases I hit in the data: stocks under $1, stocks that barely trade and sit frozen for weeks, and very small offerings. Third, I would change the timing. Question 3 showed the longer you hold, the worse the typical stock does. So rather than buying on day one and waiting, I would test buying around six months in, after insiders are free to sell and that selling has passed, with a capped holding period or a stop-loss. The catch is that betting on a few picks means a bad selection rule loses everything, not just underperforms. I would test any of this against the 2023 and 2024 IPOs first. 1
To build a more profitable IPO investing strategy and overcome the typical negative median returns of naive buy-and-hold approaches, you should shift toward a selective quantitative framework. This involves filtering out speculative assets like SPACs and shell companies in favor of companies with established revenues (e.g., >$100M) and top-tier underwriter backing. Strategically, waiting to buy until after the 180-day lock-up expiration avoids initial downward selling pressure, while employing technical trend-following momentum indicators (such as waiting for prices to break above a short-term SMA or showing relative market strength) helps ensure you do not buy a falling knife. Furthermore, focusing strictly on highly sought-after IPOs priced at the top of their expected ranges and adopting strict risk management rules—such as applying a 7% to 10% trailing stop-loss and utilizing a short-term tactical holding period of 30 to 60 days—will protect your capital and capture peak performance before returns begin to deteriorate. 1
To significantly increase the profitability of an IPO trading strategy, consider: 1. Inverting the Strategy / Systematic Short-Selling: Because the median 12-month return is -51% (growth = 0.49), shorting non-profitable IPOs or buying post-IPO put spreads aligns directly with the prevailing negative drift. 2. Fundamental & Profitability Filters: Filter out speculative shell companies, SPACs, and pre-revenue biotech. Restrict entries strictly to companies with positive EBITDA, positive Free Cash Flow, and >25% verified revenue growth. 3. Institutional & Underwriter Quality: Prioritize IPOs led by tier-1 bookrunners (Goldman Sachs, Morgan Stanley, JP Morgan) and high institutional anchor allocations (>70%), which exhibit substantially better price stabilization. 4. Lock-up Expiration Arbitrage: Avoid entering on Day 1. Instead, monitor the 90- to 180-day insider lockup expiration date; wait for the supply absorption before initiating positions. 5. Technical Momentum & Base Breakout: Only enter if the stock forms a consolidation base and breaks above its Day 1 high with surging volume, coupled with a strict 7% stop-loss. 1
Ajustaria a estratégia para comprar quando o RSI < 30 e vender após uma recuperação, testando diferentes períodos de saída para aumentar a lucratividade. 1
Instead of holding the stock for a mandatory 30 days while waiting to see what happens, the strategy should include an automatic exit rule. For example, sell immediately if the stock drops 5%, or sell if it rises 15% to lock in the profit before it falls again. 1
(1) Don't buy on day one, (2) Use data as advantage: Use technical confirmation, condition on market IPO returns are very cyclical, and Model it. 1
Filter by fundamentals. Only buy IPOs that were already profitable, with positive EBITDA margin and real revenue growth. 1
Filter by company type / sector; Use pre-IPO signals; Entry timing after IPO 1
After investigating what I observed using the available data I found out a strategy that may make sense. ### IPO Short-Term Momentum Strategy (backtest by a practitioner) Here you can find the article I read ([Article](https://www.quantitativo.com/p/the-edge-in-trading-ipos-18-annual)). You can see that the author is bothered by capital utilization, which may be a concern. * Core insight: Newly listed stocks exhibit a ~21-day momentum window * Buying at an all-time high yields ~3% expected return per trade vs ~1% non IPO stocks making new highs Considerations * The initial underpricing creates a "discovery phase" where institutional investors accumulate over the first few weeks. The all-time-high close is a confirmation that this accumulation is still in progress. * On the first trading days (~21) you can find the "sweet spot", beyond that the IPO-specific edge fades. * Asymmetric reward to risk, 2:1 proposed reward-to-risk Steps * Companies listed within the past ~90 days (mid/large/mega cap only) * (Day-1 Price * Volume ) > 10M in order to avoid liquid names * Buy at the open on the day after the stock closes at a new all-time high, deliberately skipping the initial volatility (days 1-2) * Profit target +20% (limit order placed same day as entry) * Stop-loss -10% (stop order placed same day as entry) * Max positions 20 * If more than 20 signals fire per day, prioritize market cap (largest first) * Replace when a target or stop is hit, the freed slot is filled by the next qualifying signal 1
Profitability could potentially be improved by combining RSI with trend and volume indicators, such as the 200-day SMA, to strengthen trade signals and reduce reliance on a single indicator. Given the observed negative average and median 12-month IPO returns, downside strategies such as short positions or put options could also be investigated, particularly after the initial post-listing period. Fundamental filters, including profitability, revenue growth, cash flow, valuation, and company structure, could further help identify stronger IPO candidates while distinguishing traditional IPOs from SPACs and other acquisition vehicles. 1
Incorporating a good selection stock filter. It involves to not select random companies, but companies with a good liquidity, solvency, profitability, growth projections, good investment projects and more. A good input will return a good output 1
6-filter quantitative framework 1
1. Filter Before Buying Only invest in IPOs that are profitable at listing, have strong revenue growth, and were backed by top-tier underwriters. Skip hype-driven, pre-revenue names. , 2. Don't Buy on Day 1 Wait roughly 6 months (past lockup expiration) before entering, since IPO-day prices are typically inflated by initial hype., 3. Diversify Hold a basket of top-ranked IPOs rather than concentrating in one or two, since IPO returns are right-skewed (a few big winners offset many losers). 1
- Implement Quality and Fundamental Filters (Screening) - Short-Selling or Hedging (Betting Against Hype) - Post-Lockup Expiration Entry - Utilize Technical Momentum Indicators - The 'Underpricing Flip' (Primary Market Allocation) 1
To significantly enhance profitability, reduce drawdown, and eliminate "falling knife" drag from the baseline RSI < 30 strategy, implement the following four quantitative improvements: Trend & Regime Filtering (Avoid Structural Downtrends): Issue: Buying purely on RSI < 30 often forces entries into financially distressed companies in secular decline. Improvement: Require macro trend alignment — only accept an RSI < 30 buy signal if the stock price is trading above its 200-day Simple Moving Average (SMA) or if market-wide regime indicators (e.g., S&P 500 above 50-day SMA) are bullish. Signal Reversal Confirmation: Issue: Buying while RSI is actively falling risks entering midway through an ongoing sell-off. Improvement: Instead of buying when RSI crosses below 30, trigger the buy order only when RSI crosses back above 30 (or above 35), confirming that selling pressure has dried up and bullish momentum is resuming. Dynamic Risk-Defined Exit Rules (Stop-Loss & Take-Profit): Issue: Holding every trade for a rigid 30-day fixed duration leaves open profit vulnerable to market reversals and uncapped downside. Improvement: Set an ATR-based Stop-Loss (e.g., 2 i m e s e x t A T R ) to exit losing trades immediately. Set a Dynamic Profit Target (e.g., sell when RSI reaches 60–70 or when profit exceeds + 8 ). Volatility-Adjusted Position Sizing: Issue: Fixed $1,000 allocation treats low-volatility blue chips and hyper-volatile penny stocks identically. Improvement: Scale position size inversely to stock volatility (Inverse-Volatility Sizing), or scale entries deeper into oversold territory (e.g., $1,000 at RSI 30, additional $1,500 if RSI touches extreme oversold < 20). 1
1) Screen for positive EBITDA / GAAP net income and institutional backing; 2) Avoid entering day 1 close blindly; enter only if the stock consolidates in an "IPO base" and breaks out. 3) Enforce tight risk limits (-7% stop-loss) and exit before day 90/180 lockup expirations. 1
I would add filter layers to ensure higher-quality trades: 1) Only invest in IPOs with positive trailing revenue, 2) Restrict trades to top-tier underwriters, and 3) Implement a dynamic stop-loss to protect against immediate post-IPO decay. 1
I would replace the unconditional strategy of investing in every IPO with a selective, prediction-based strategy. I would first remove offerings that are not comparable to traditional IPOs, such as SPACs, direct listings, uplistings, and securities with insufficient liquidity. I would then create features available at the investment date, including offer size, initial valuation, revenue growth, profitability, sector, underwriter, first-day return, trading volume, market volatility, sector momentum, and overall market regime. 1
Risk Management: Use the SMA 20 as a trailing Stop-Loss to exit immediately if the stock breaks back down, cutting losses short while capturing the rare high-growth winners. 1
I would make the strategy more selective by combining: - Positive recent momentum. - Lower volatility and sufficient liquidity. - Performance relative to the market or sector. - Position limits, transaction costs, and stop-loss rules. - Out-of-sample testing using only information known before each trade 1
To increase IPO profitability, first filter out SPACs and acquisition corps (which Q1 shows have the highest withdrawn value) and require positive revenue growth, institutional ownership above 30%, and a minimum $100M float; then apply valuation discipline by only buying at or below 20x forward revenue with a discount to public comps and avoiding deals priced above the original range midpoint; for timing, wait until after the 180-day lock-up expires and the first earnings report (since Q3 shows median growth deteriorates the longer you hold, with 1 month being the least negative at 0.94), and only enter if the stock trades above its 20-day SMA with RSI above 50 (Q4's 55% win rate confirms avoiding oversold traps) on surging volume; overweight secular growth sectors like AI and GLP-1 pharma while using macro indicators (Fed funds, 10Y yield, CPI) to time entries, and enforce strict risk management with 2% position limits, 15% stop-losses, and scaled entries across the lock-up and earnings milestones — this focused approach should shrink the universe to 15-20 high-conviction names and shift median returns from -6% to positive territory. 1
Filter IPOs by sector, focusing exclusively on growth industries (technology, biotech) while excluding defensive sectors that historically underperform. Only execute IPO trades when the broader market (S&P 500, Nasdaq) is in an uptrend, indicated by positive 30-day returns, to avoid fighting headwinds from market downturns. Avoid holding IPOs through their lock-up expiration periods (typically 180 days), as these dates frequently trigger sharp selloffs. Either exit before lock-up ends or short the stock 30 days prior to capitalize on the expected decline. 1
Categorize the IPOs by sub-sector and only take trades in sectors where the broader market ETF is trading above its 50-day moving average. 1
Don't buy every IPO at the first close. Filter with a classifier trained on features known at listing (deal size, pricing vs. range, first-day pop, underwriter tier, sector, SPAC flag, VIX/S&P momentum), avoid tiny $4 micro-cap deals, and enter later (after the quiet period or lock-up expiry). Add stop-loss, take-profit and trend filters (price above SMA50/200), consider shorting weak IPOs since the median drifts down, and validate with a walk-forward backtest including costs. 1
I would buy selectively: favor IPOs with strong revenue growth, reasonable valuations, and positive price momentum after the first month. I would also consider market conditions, diversify positions, and test different exit rules. To verify improvement, I would test on newer IPOs excluded from strategy development, including trading costs and delisted stocks. 1
I would not buy every IPO; I would drop acquisition/blank-check names and keep only names that are up after the first month. 1
I would add filters based on valuation, trading volume, market conditions, sector performance, and early price momentum. I would also test different holding periods and risk-management rules, and compare the strategies using Sharpe ratio, maximum drawdown, and win rate to identify approaches with better risk-adjusted returns. 1
Do not buy every new listing and hold it. Median growth from the first close is below 1 at every horizon from 1 to 12 months, and it gets worse the longer you hold, so a fixed buy-and-hold is a losing typical trade. I would: Drop blank-check / zero-return names and other listings that are not really trading. If buying near the IPO, exit on a short horizon (the median multiple is highest at 1 month). Only take the trade when simple filters point to a better base rate: a real first-day close above the offer, a known underwriter, enough size and liquidity, and a market that is not already falling. Otherwise skip the IPO and wait for an oversold RSI signal. That rule had a positive average 30-day return on the long history, which the unconditional IPO hold does not. 1
Avoid buying at the end of the first trading day, as the median return from that price is negative across all holding periods from 1 to 12 months. Instead, consider waiting until the lock-up period expires (approximately 90–180 days) or until the price stabilizes. Select IPOs carefully before investing by evaluating deal size, underwriter reputation, SCOOP ratings, and whether the offering price was set above the proposed price range. Avoid SPACs and micro-cap stocks. Enter a position only when there is momentum confirmation (the price is above the SMA20/50 or the RSI indicates oversold conditions) and market conditions are bullish (the S&P 500 is above its SMA200 and the VIX is low). Set stop-loss and take-profit levels, and diversify across multiple IPOs, as most gains tend to come from a small number of stocks that experience extreme price increases. Finally, train an ML model (such as gradient boosting) to predict the probability of positive returns, then trade only IPOs with a high predicted probability of success. 1
1. Filter before trading. Only buy: NYSE large raises ($50M–$500M), tier-1 underwriters (GS, MS, JPM), non-SPACs, non-uplistings. Skip micro-caps and wide price bands. 2. Better entry timing. Not day 1. Wait for the quiet-period expiry (day 22–25) or after lock-up (180d). Enter only if rsi<40, SMA10>SMA20, macdhist>0. 3. Regime gate. Only trade when SPX 30d growth > 0 and yield curve is upward-sloping. Sit out risk-off. 4. Smarter exits. Use 2×ATR trailing stop, take profit on half at +50%, hard stop at −10%. Never a fixed 30-day exit. 5. Size asymmetrically. Scale by inverse vol (parquet has volatility); concentrate capital on top-ranked signals, skip the tail. 6. De-duplicate. Consecutive RSI<30 days = one position, not many. 7. Portfolio limits. Max 5 concurrent positions, 20% per sector. 8. Mirror trades. Short day-1 pops >40%, short into lock-up expiry — these often have positive medians. 1
I would be more defensive when trading stocks that have recently IPO'd given their track record. Maybe I could use an RSI indicator to flag when an IPO stock is oversold so that I would know that it's been discounted from the initial price. Even then, I would further narrow down the stocks to does that I know has an established moat and I've done my research on. 1
Shorten the holding period. Question 3 showed the median growth peaks at month 1 and declines afterward — holding longer destroys value rather than adding it. A "buy at IPO, exit within ~1 month" strategy would capture the initial pop without riding the later decline. Filter for quality before buying, instead of taking every IPO. Question 1's data was full of SPACs ("Acquisition Corp" / "Blank Check" companies) — these vehicles are historically associated with worse returns than operating companies going public. Excluding SPACs from the universe would already remove a lot of noise. Select for Sharpe ratio, not raw growth. The mean Growth_252d was ~1.76x the median — a handful of extreme winners skew the average while most IPOs underperform. Instead of trying to pick "the next big winner," a Sharpe-based selection favors stocks with more consistent risk-adjusted returns, reducing dependence on guessing the outliers correctly. Diversify like a venture-capital portfolio. Since returns are concentrated in a small number of IPOs (same skew visible in the 25%/75% spread from .describe()), betting on a single IPO is close to a lottery ticket. Spreading small amounts of capital across many IPOs increases the odds of capturing the few that take off, without needing to pick winners individually. Combine with a technical signal (like the RSI from Question 4). Instead of buying on the IPO day itself, waiting for a short-term pullback (low RSI) after the debut could improve entry timing — the same "buy the dip" logic already applied elsewhere in the course. Pick 2–3 of these and write 3–4 sentences — it's a reflection question, doesn't need to be exhaustive. 1
Filter out micro-cap IPOs. Don't buy at the first-day close; wait and confirm momentum. Avoid IPOs that jump more than 50% on day one. Time exits around known events. Add risk management 1
Wait for the post-IPO correction instead of buying at the peak of the hype. Our Q3 results show that median returns are highest in the first month and decline afterward, meaning prices tend to drift down as the initial excitement fades. A better entry point could be after the lock-up period expires (typically 90–180 days after the IPO), when insiders are allowed to sell and the resulting selling pressure often pushes the price lower. Buying after this pullback lowers the entry price and filters out stocks that only rose on hype. Or build a dataset of past IPOs with features such as sector, offering size, where the final IPO price landed relative to the initial price range, first-day return, early volatility, and overall market conditions (e.g., whether the S&P 500 is above its 200-day moving average). Then train a classifier to predict whether the stock's growth after N months will be greater than 1 (i.e., a positive return). The strategy would only buy IPOs where the model's predicted probability of a positive return is high, filtering out the majority of IPOs that tend to decline. 1
Considering the data from Questions 2–3, buying every IPO at its first close would lose money for the typical stock: after 12 months the median 2025 IPO is at 0.55 of its first-day close, and only 28 % of the stocks are up. So we could increase profitability by selecting the IPOs and timing the trades, with these strategies: Skip the small and cheap deals (deal size below $50 M or offer price $5 or less). Wait for an entry signal instead of buying at the first close. Predict the winners with a model. Cut the losers and keep the winners. These ideas would still need a backtest against the S&P 500 after fees. The real results would likely be worse, since 14 of the 146 tickers are missing from Yahoo Finance (probably delisted). 1
add indicator Divergence - Convergence 1
I would make the strategy more selective by: Using a lower RSI threshold (e.g., RSI < 25) to identify stronger oversold conditions. Adding a trend filter, such as only buying when the stock is above its 200-day moving average. Testing different holding periods to find the period with the highest historical returns. Adding stop-loss and take-profit rules to control downside and lock in gains. This could reduce the number of trades but potentially improve the profitability per trade. 1
Most IPOs lose money, so I would stop buying all of them. I would only buy IPOs that closed above their offer price on day one or have a good SCOOP rating, and skip tiny, very volatile ones. I would also sell early, since the median return was best after one month and got worse after that. Finally, I would spread small amounts across many IPOs, because a few big winners make most of the profit. 1
My results show that buying a 2025 IPO at its first close and holding it is a losing strategy for the typical stock. The median growth was about 0.93 after 1 month and fell to about 0.53 after 12 months, while the mean was inflated by a few huge outliers (one stock grew about 33x). So I would change the strategy in four ways. 1. Change the entry. Don't buy at the first close, when the IPO pop is already priced in. Wait for the early selloff, or for the lock-up expiry (around 180 days), and enter only if the stock has stabilized. 2. Filter what to buy. Use information available at IPO time, such as offer price versus the filed range, deal size, sector, underwriter quality, first-day return, and market conditions (VIX, recent IPO volume). Exclude SPACs and micro-caps, which drag the median down. 3. Screen on risk-adjusted returns. After 60 to 90 days of trading history, keep only stocks with a positive Sharpe ratio and moderate volatility, since the Sharpe median (0.048) was more informative than raw growth. 4. Manage risk. Use small position sizes, stop-losses, and diversification across many IPOs, since a handful of winners drive the average. I would also cut losers early, given how quickly the median decays. Finally, I would train a simple model (for example gradient boosting) on these features to predict positive 30-day or 90-day returns, validate it with a time-based split rather than a random one, and include transaction costs so the result isn't overfit to 2025. 1
Rather than buying every IPO at its offer price, use a selective, rules-based approach: - Screen for durable fundamentals: revenue growth, improving gross margins, manageable debt, positive or improving free cash flow, and enough cash runway. - Compare valuation with established peers and avoid offerings priced for unusually optimistic growth. - Treat float, lock-up expiry, insider ownership, underwriting quality, and institutional demand as risk factors, not standalone buy signals. - Wait for price discovery (for example, require a base and a confirmed trend after the first few weeks) instead of assuming the first-day close is an attractive entry. - Test explicit entry and exit rules with walk-forward, time-based out-of-sample validation; include slippage, fees, delistings, and a broad-market benchmark to avoid survivorship and look-ahead bias. - Limit position size and diversify across sectors and IPO cohorts; define a maximum loss and an exit horizon before entering. 1
**1. Manual (rule-based) filters.** Screen deals before entry using observable characteristics: skip SPACs/blank-check companies, prefer large offerings (>$300–500M) with reputable lead underwriters, require signs of strong demand at pricing (priced above the initial range, upsized deal), and check fundamentals in the S-1 filing (revenue growth, path to profitability). Even simple heuristics like these should cut out the long left tail of low-quality IPOs that drives the negative median return. **2. Binary classification model.** Frame it as a supervised learning task: predict whether an IPO will have a positive return over a fixed horizon (e.g., 3–12 months) using features such as deal size, underwriter tier, industry, pricing vs. range, first-day return, and market regime. Because returns are right-skewed — a few big winners drive all the profit — the model should be optimized for precision in the top predictions, not overall accuracy: the goal is to identify the small subset of IPOs worth trading, not to classify every deal correctly. 1
If there is high probability that the stock price will go down... then we should buy put options... or sell call options 1
I would refine the strategy by filtering stocks based on factors such as sector, company size, IPO valuation, revenue growth, profitability, trading volume, and post-IPO price performance. I would then compare these features with similar established companies and backtest the strategy on historical IPO data to identify which characteristics are associated with stronger performance. 1
A better strategy is to build a predictive classification model that estimates the probability that a newly IPO’d stock will have a positive return over the chosen holding period. 1
I would test filters available at the time of purchase—for example, offering size, valuation, sector, first-day price movement, and liquidity—then compare them on a later IPO cohort. I would include delisted stocks and trading costs, and judge the strategy by its median return and downside as well as its mean. 1
I would filter for signals of positive future returns: pre-listing demand (grey-market premium, oversubscription), fundamentals (revenue and profit growth, valuation vs listed peers), and market context (sentiment, sector momentum, events near the listing). 1
I would not allow auto investment strategy and apply a better filter based on company characteristics, liquidity, and post-IPO momentum. Besides, I would also implement dynamic entrys and exits based on rules such as trend indicators or other mechanisms. 1
To turn negative IPO returns into a profitable strategy, avoid buying on Day 1 when hype and valuations peak. Instead, wait 15 to 30 days after the 180-day lockup period expires, allowing insider selling pressure to subside and the stock to establish a realistic price floor. Combine this timing with strict quality filters—selecting only companies backed by tier-1 underwriters with positive gross margins and strong revenue growth—and enter the trade only when price action confirms an uptrend above its 20-day moving average. 1
I would avoid buying every IPO indiscriminately and add filters such as positive price momentum, trading volume, company fundamentals, and market conditions. I would also use risk management such as position sizing, stop-loss rules, and backtesting different entry and exit conditions to improve risk-adjusted profitability 1
Several ways i can change: - Filter by quality: Only trade IPOs that are profitable at offering, lead-managed by bulge-bracket banks, with high SCOOP ratings - Don't buy on day 1: Wait for a pullback (e.g., buy after -20% drawdown or when RSI crosses back above 30) - Add technical filters: Combine RSI < 30 with price above 200-day MA, MACD bullish cross, etc. - Risk management: Use trailing stops (e.g., 2×ATR), hard stop at -15%, take-profit at +30% - Position sizing: Risk only 1% of equity per signal, max 10 concurrent positions - Avoid lock-up periods: Skip stocks within 30 days of lock-up expiry - Macro filter: Skip signals during high-rate/recession regimes using FEDFUNDS, CPI data - ML ranking model: Train a classifier on historical IPOs to predict P(positive return) using fundamentals, underwriter, sector, market regime 1
As explored in the linked notebook, I first tested replacing the automatic first-day entry with an entry at each IPO’s first oversold episode (RSI < 30). For the same 123 IPOs, this improved the median return and win rate, but reduced the mean return from 4.96% to 1.27%. Therefore, RSI alone did not increase profitability. To directly target profit, I would use RSI and its state dynamics as features alongside first-day performance, momentum, volatility, volume, IPO characteristics, and the broader market regime. A model would be trained with a temporal split to estimate the expected future return of each IPO, and capital would be allocated only to the highest-ranked opportunities. The optimization target should be expected return or net profit rather than classification accuracy alone, while future-return columns must be used only as targets to prevent data leakage. 1
To enhance IPO trading profitability, strategies can incorporate a combination of fundamental quality checks, refined entry timing, and broader market filters. 1
I would be more selective about IPOs, looking at revenue growth, debt and valuation. I would also test waiting for the price to settle before buying. I would develop these rules using older IPOs, then test them on later IPOs, including trading costs and delisted companies. 1
Instead of a rigid threshold, use historical IPO data to train a straightforward classification model (like Random Forest or Logistic Regression). This allows the system to weigh multiple variables and predict the actual probability of a positive return before risking capital. 1
The naive "buy-at-first-close" rule fails because IPO returns are skewed (the median sits below the mean), the first-day pop is already priced into the entry, and lock-up expiries create selling pressure. To improve performance, trade only top-scoring IPOs (filtering by offer size, price revision vs. range, underwriter reputation, and first-day volume), enter after the post-IPO pullback instead of at the first close, and implement stop-loss and time-based exits. Additionally, incorporate a market-regime filter (such as an index above its 200-day moving average and a low VIX), hedge market beta with an index or sector ETF to isolate true IPO alpha, and validate the strategy using a walk-forward split and realistic transaction costs, given the small and survivor-biased sample size. 1
For this, we probably need to close during the first day, and look for the optimal time for such closing on the position. Also, we can add more information to decide which IPOs are worth investing in. Finally, other approaches such as fundamental analysis, machine learning models, and such can help in creating stronger indicators. 1
To transform the negative baseline performance of IPO investing into a profitable system, I would shift from a passive "buy-all" approach to a highly selective predictive framework: 1. Implement a Lock-Up & Wait Filter: Filter out entries during the first 90–180 trading days to avoid the typical post-IPO decay and insider liquidation pressure. 2. Probability-Threshold Model Evaluation: Train a Machine Learning classifier (e.g., XGBoost) on precomputed indicators (like ADX, RSI, and SMA trends) alongside macro data (like FEDFUNDS and CPI changes). I would restrict entries to high-conviction trades where the model's probability score exceeds 75%. 3. Fundamental Conditioning: Enrich the dataset with fundamental attributes, programmatically requiring that an IPO must exhibit strong positive revenue growth or top-tier institutional underwriting to be eligible for entry. 1
Change the entry point, filter which IPOs I would like to buy, change size and exit plan for me, build a predictive model, change the benchmark 1

Calculated: 30 September 2026, 16:13