"Buy support" is one of the most repeated pieces of advice in trading, and one of the vaguest. Support is a level where buyers have shown up before. The trouble is that on a chart you can draw one almost anywhere, and once it is drawn your eye will find the times it held and skip the times price went straight through it.
This guide takes the idea and makes it testable. We ran it on Trade Manager on 7 September 2026 across every stock in the S&P 500. It produced 48 trades and won 54.2% of them. The more useful part is why, and it is not the part most people would guess.
Turning "support" into a rule
A computer cannot see a hand-drawn zone. It needs a definition, so we used two conditions, both on the daily timeframe:
- Trigger — a 100-bar low. Price reaching the lowest point of the last hundred bars. This is deliberately strict. It fires rarely, and when it does the level is genuinely significant rather than a minor dip inside a range.
- Confirmation — RSI(14) below 30. Under 30 means the last fourteen bars have been overwhelmingly red. It is not a prediction, just a statement that selling pressure is running out.
- Exits — a 5% target and a 3% stop.
The reason both conditions are needed is that either one alone is a bad filter. A 100-bar low on its own catches falling knives. An RSI under 30 on its own fires constantly in any downtrend. Together they describe something narrower: price at a level that has mattered before, reached by selling that has largely run its course.
The result
| Metric | Value |
|---|---|
| Closed trades | 48 |
| Won | 26 |
| Lost | 22 |
| Win rate | 54.2% |
| Maximum drawdown | −14.1% |
| Average hold | 6 bars |
| Universe and window | 503 of 503 tickers, 220 calendar days |
Real trades from the run, winners and losers both: NFLX stopped out at −3%, ORLY hit target at +5%, PYPL stopped out, MA hit target, BRK.B hit target, APO hit target.
We are deliberately not putting the cumulative return in a heading. It was positive, and it is reported inside the app, but a compounded figure across 48 trades over ten months reads like a promise when it is lifted out of context — and the caveats at the bottom of this page mean the real-world number would be lower. The win rate and the trade count are the honest summary.
The number that decided it
A 54.2% win rate is only meaningful next to the number the exits demand:
Break-even win rate = stop ÷ (target + stop)
With a 5% target and a 3% stop, that is 3 ÷ (5 + 3) = 37.5%. The strategy scored 54.2%, clearing the requirement by 16.7 percentage points. That margin — not the cleverness of the setup — is what made it work.
The comparison worth making is with the hammer candle test, run on the same engine and the same index:
| Support bounce | Hammer | |
|---|---|---|
| Exits | +5% / −3% | +4% / −3% |
| Break-even needed | 37.5% | 42.9% |
| Win rate achieved | 54.2% | 38.7% |
| Margin | +16.7 | −4.2 |
| Outcome | Positive | −32.7% |
The hammer needed a higher win rate than the support bounce did, purely because its target was smaller relative to its stop — and it delivered a lower one. Two strategies, the same market, the same engine, and the thing that separated them was arithmetic set before either took a trade.
What a 54.2% win rate does not mean
It does not mean the next 48 trades will look like these. Three things are worth holding on to before anyone treats this as a system:
- 48 trades is a small sample. Enough to be interesting, not enough to be conclusive. A different ten-month window could easily produce 45%.
- The window was one market regime. Roughly ten months of a particular market. Mean-reversion setups behave very differently in a sustained downtrend, where every 100-bar low is followed by another one.
- A −14.1% drawdown is the part you would have had to sit through. The final number is easy to look at afterwards. Being 14% down while following the rules is the actual test.
Run it yourself
- Open the strategy builder and choose a universe.
- Add a Support/Resistance block as the trigger, set to a 100-bar low.
- Add an RSI block as confirmation, below 30.
- Set your target and stop — then work out stop ÷ (target + stop) and write it down before running anything.
- Run the backtest and compare its win rate to your number. That comparison is the result.
Backtests replay bar by bar using the same evaluation code as the live scanner, so a strategy that tests one way behaves the same way once alerts are switched on. If it clears your break-even number with margin, you can activate it and get an alert the moment it triggers instead of watching charts. See how to backtest a trading strategy without coding for the mechanics.
The caveats, which both flatter this result
Survivorship bias. The test uses index membership as it stands today. Companies dropped from the S&P 500 during the period are absent, and removals skew toward poor performers.
Frictionless fills. Trades are assumed to fill exactly at the target or stop, with no slippage and no commissions.
Both push a positive result upward. That matters more here than it did in the hammer test: a losing strategy under kind assumptions is definitely losing, but a winning strategy under kind assumptions is only probably winning. Treat 54.2% as an optimistic upper bound.