Win rate is the highest-leverage number in your whole plan. As we showed in the compounding article, nudging it from 58% to 64% changes the five-year curve dramatically — because every extra winner compounds. The catch is that you do not raise win rate by predicting better. You raise it by being more selective: trading only the setups where the odds are genuinely tilted, and passing on everything else.
Three filters do most of that work, in order: score the technicals, read the sentiment, and skip the earnings. Each one removes a category of trade the edge cannot survive.
Filter 1: score the technical setup — don't eyeball it
Eyeballing a chart is where win rate quietly leaks away, because a discretionary read is inconsistent. The fix is to score every setup the same way and only trade the high-quality ones. Our 0–100 opportunity score blends three pillars — gamma corridor, IV regime, and trend conviction — and maps to a decision tier: APPROVED at 65+, watchlist from 40–64, no-trade below 40. (The full mechanism is in how we automate technical analysis into a score.)
The win-rate lever here is simple: only take APPROVED setups, and let the conditions choose the structure. Most candidates will not clear the bar on any given day — and that is the point. Fewer, cleaner trades win more often than more, sloppier ones.
Filter 2: read the sentiment — the fundamental overlay
A setup can be technically clean and still sit on top of a fundamental landmine. A sentiment layer catches what price has not yet reflected. Our composite blends four independent reads — news and analyst coverage, X / social chatter, sector and macro narratives, and retail forums — into a single 0–100 sentiment score (bullish 70+, neutral 40–69, bearish below 40), each with its own confidence weight.
Crucially, sentiment is an overlay, not the thesis. It acts in three ways:
- Confirm (boost): when sentiment is strong (≈75+) and agrees with the technical direction, the setup gets a small score bump.
- Diverge (caution): when sentiment contradicts the technicals — bearish chatter under a bullish chart — the setup is flagged and marked down, not blindly traded.
- Suppress (veto): when a hard catalyst is detected — earnings, M&A, a regulatory or FDA event — the name is pulled from consideration entirely until it clears.
Filter 3: skip the earnings — the single biggest win-rate killer
No technical edge survives an earnings report. A stock can gap 5–15% on a surprise, blowing straight through short strikes that looked perfectly safe the day before — and the post-earnings IV crush punishes long-volatility positions just as hard. It is binary gap risk: a coin flip glued to the front of your trade.
So the rule is not "be careful around earnings" — it is a hard exclusion. If an earnings date falls inside the trade's window (a ~21-day default), the name is removed from the pool. Not downgraded, not sized smaller — removed. A related rule handles ex-dividend dates for any structure with a short call, where early-assignment risk spikes.
| Event in the expiry window | Action |
|---|---|
| Earnings report | Excluded — no trade until it passes |
| Ex-dividend (short-call structures) | Excluded — early-assignment risk |
| No binary events | Eligible — the edge can play out |
This one filter does more for realized win rate than almost anything else, because it deletes the trades most likely to produce a large, sudden loss.
Stacking the filters: fewer trades, higher hit rate
Run all three gates and the candidate list collapses — by design. What survives is a small set of trades that are technically scored, sentiment-clear, and free of binary events before expiry. That is how a defined-risk premium-selling plan realistically targets a 60–70% win rate: not by being clairvoyant, but by refusing the trades that drag the average down.
And remember why this matters so much. A few points of win rate, applied across hundreds of compounding trades, is the difference between a curve that drifts and one that triples. Selectivity is not caution for its own sake — it is the cheapest edge you can buy.