Video summary
Former Poker Player: The Exact Math That Helped My Trading!
Main summary
Key takeaways
Finance-focused summary (Sep 3, 2026 interview)
Who / where
- The video is an interview within the CMR community (with a helpdesk mentioned).
- Jeremiah discusses trading performance using probabilistic/statistical “edge” concepts drawn from poker and gambling markets.
- Presenters:
- Jason Shapiro (host)
- Jeremiah (guest)
Core finance / investing ideas & key recommendations
Emphasize EV and payout structure (not direction alone)
- Jeremiah stresses evaluating trades by:
- win rate
- expected value (EV)
- payout structure
- The goal is to avoid focusing only on whether the market goes up or down.
Manage downside tightly
- He controls risk such that he is typically risking below 1% per trade about 99% of the time.
- He also acknowledges occasional larger mistakes:
- sometimes risking around 100 basis points (bps) per trade.
Use stats to diagnose negative performance, then adjust incrementally
- If results are negative, use trade statistics to identify why (not just accept the outcome).
- Make small, incremental changes rather than sweeping redesigns.
Avoid over-trading (timing/regime matters)
- He suggests that sometimes “now is not the best time to trade.”
- This implies that:
- trade frequency and
- market regime/timing can strongly affect outcomes.
Prefer lower variability (lower standard deviation)
- He contrasts:
- high-variability approaches (e.g., some trend-following behavior) vs.
- smoother performance
- His preference: lower standard deviation / lower variability.
Jeremiah’s step-by-step framework (turning stats into an improved system)
1) Collect clean performance data
- Keep accounts separate for clean datasets.
- He references tracking via:
- FundSeeder connected to IBKR
- He also advocates separating accounts by strategy.
2) Compute and interpret metrics
Track (at minimum):
- % successful trades
- average win
- average loss
- expected value per trade
- (Implicitly) variance / standard deviation
3) Identify the binding constraint (what is limiting EV?)
If overall EV is negative, determine what’s “causing” it:
- Too-low win rate
- Improve trade selection (better odds of winning)
- Make fewer/better trades
- Wins are too small / exits mistimed
- Consider whether you’re:
- selling too soon in uptrends
- selling too late in downtrends
- Consider whether you’re:
- Losses are too large
- Consider narrowing stops
- Exit losing positions faster
4) Change one variable at a time
- Use small changes.
- Ideally adjust only one parameter per test so you can attribute cause more confidently.
5) Validate using larger samples
- He emphasizes that systems that behave like “coin flips” require large sample sizes for statistical reliability.
6) Use a trading journal
- Journal every trade so you can analyze patterns over time.
- Focus on whether categories/themes work consistently—not just on remembered winners.
Key numbers & performance metrics cited (stocks)
Trade quality / distribution
- Success rate (stocks): just over 30%
- Average win (stocks): 7.4%
- Average loss (stocks): 3.59%
EV outcome (negative overall expectation)
- Even though wins are larger than losses, the overall result is still negative EV.
- He describes this as roughly ~4 bps loss per trade (EV ≈ -4 bps per trade).
Time horizon & year-by-year results
- 18-month stock trading period
- EV: -4 bps per trade
- Approx. performance over ~200 deals: about -8%
- Breakdown mentioned:
- Prior year: +3%
- 2026: about -10%
- Net: about -8%
Futures comparison (mentioned, not fully aligned dataset)
- He says futures success rate is very similar
- Futures average profit is higher
- Main takeaway: stocks still show negative expectation
Illustrative counterfactual improvement
- If average loss could be reduced from 3.59% → 2.5%, holding other metrics constant:
- EV rises to +67 bps per trade
- Over 250 trades: described as ~+17% profit
- He attributes the improvement specifically to reducing average loss, not increasing win rate.
Trading mechanics / implementation details mentioned
Charting & execution style
- Uses daily charts for evaluation.
- Executes with real order types:
- Stop-limit for entry
- Stop-market for exit
- Limit orders for take-profit
Position sizing & risk
- Typical risk: < 1%
- He admits occasional larger bets (around 100 bps per trade), contributing to drawdowns (notably referenced around March and July).
Macro / market discussion
- He argues against outcome-chasing narratives like “predict what the S&P will do tomorrow.”
- Even if someone is “right” 50% of the time, that may not indicate real knowledge—it may just reflect coin-flip odds.
- Strategy implication:
- optimize EV/risk-reward and execution discipline
- avoid relying on directional prediction.
Instruments / tools / examples mentioned
- S&P (referenced generically)
- Polymarket (prediction markets)
- American Idol prediction market (used as a forecasting/edge example; not treated as a financial instrument)
- IBKR (broker platform)
- FundSeeder (analytics / tracking)
Disclosures / cautions (as implied by the subtitles)
- No explicit “not financial advice” disclaimer was present in the provided subtitles.
- Implicit cautions include:
- Results depend on market conditions and metrics can change by year (e.g., strategies that worked in 2022 may fail in 2026).
- Strategy changes are interconnected—adjusting one metric can affect others.
- His analysis is based on his own data; outcomes are not guaranteed.
Presenters / sources mentioned (end of video)
- Jason Shapiro (host)
- Jeremiah (CMR participant/guest)
- Hazem (credited as a creator of CMR; last name not provided)
- Nate Silver (analogy to statistics/polling via 538)
- Wikipedia (referenced for “Dial Idol” page)
- Freakonomics (article about Dial Idol in 2008)
- Dial Idol (software / past dataset example)
- FundSeeder and IBKR (tools referenced for performance tracking)