Here is the one to pay attention to. Across this entire research series — dozens of strategy/symbol combinations, most of them losing clearly after costs — the previous-day high/low break is the single closest result to breakeven we have published. On ETH its profit factor came in at roughly 1.00, meaning gross profit and gross loss were almost exactly equal: the rule paid for itself and very nearly paid for its own fees on top. That does not make it a winner — it still finished marginally negative on both symbols — but in a dataset where baselines typically lose badly, "marginally negative" is the standout, and it is worth understanding why.
The setup is simplicity itself: mark yesterday’s high and yesterday’s low, then trade the first break of either — long above the prior-day high, short below the prior-day low — with a structure stop and a 2R target. We ran it on hourly candles from 12 months of real Binance 1-minute data on BTC and ETH, both directions, 1% risk, 0.05% commission per side, on the same deterministic engine that powers Secuora’s AI backtester. What makes it different from the other breakouts in this batch is restraint: prior-day levels are major, infrequent references, so the rule traded only dozens of times per symbol all year rather than hundreds or thousands — and that low frequency is most of why the fee line never buried it. Full numbers below.
Verified Result
Weak: the mechanical version barely cleared, or failed to clear, the fee hurdle.
| Market | TF | Trades | Win | PF | Max DD | Net |
|---|---|---|---|---|---|---|
| ETH | 1h | 76 | 34.2% | 1.00 | 12.1% | -0.2% |
| BTC | 1h | 69 | 33.3% | 0.93 | 13.4% | -3.5% |
How the SVS 45 breaks down ▾
12 months of real 1-minute data, fees on (0.05%/side), $10k start, 1% risk. How the score works →
The exact rules we tested
- At the start of each day, mark the previous day’s high and previous day’s low as fixed levels.
- Go long on the first hourly close above the previous-day high; go short on the first hourly close below the previous-day low.
- Cross semantics — each level arms once per day; staying beyond it does not re-signal.
- Stop beyond the last opposing swing (structure stop, fractal lookback 3); target 2R.
- No intraday session filter — crypto trades 24/7, so the prior-day levels are the only timing reference.
- Risk 1% of equity per trade; 0.05% commission per side; 10× max notional leverage.
Results
Binance spot 1-minute klines (data-api.binance.vision), aggregated per strategy timeframe · starting balance $10,000 · risk 1%/trade · Commission 0.05% per side; no spread/slippage modeled (BTC/ETH spot spreads are sub-basis-point); position size capped at 10× notional leverage. Generated 2026-06-12 by the Secuora ai-strategy deterministic runner (same engine as the in-app AI backtester).
| Month | Trades | Win rate | Net P&L |
|---|---|---|---|
| 2025-06 | 6 | 50% | $264 |
| 2025-07 | 2 | 100% | $406 |
| 2025-08 | 3 | 0% | −$328 |
| 2025-09 | 3 | 33% | −$30 |
| 2025-10 | 9 | 33% | −$48 |
| 2025-11 | 8 | 38% | $44 |
| 2025-12 | 11 | 9% | −$848 |
| 2026-01 | 8 | 50% | $332 |
| 2026-02 | 4 | 50% | $184 |
| 2026-03 | 2 | 0% | −$206 |
| 2026-04 | 4 | 50% | $166 |
| 2026-05 | 9 | 22% | −$282 |
| Month | Trades | Win rate | Net P&L |
|---|---|---|---|
| 2025-06 | 7 | 29% | −$125 |
| 2025-07 | 7 | 43% | $171 |
| 2025-08 | 7 | 57% | $481 |
| 2025-09 | 2 | 0% | −$214 |
| 2025-10 | 3 | 67% | $301 |
| 2025-11 | 7 | 29% | −$127 |
| 2025-12 | 6 | 17% | −$327 |
| 2026-01 | 12 | 25% | −$362 |
| 2026-02 | 8 | 50% | $371 |
| 2026-03 | 10 | 20% | −$436 |
| 2026-04 | 4 | 50% | $176 |
| 2026-05 | 3 | 33% | $69 |
Assumptions (how loose terms were pinned down)
- Break of the previous-day high (long) or low (short); structure stop; 2R
Why this one came closest: real levels, few trades, light fees
Two things separated the previous-day break from the breakouts that bled. First, the levels are real. Yesterday’s high and low are watched by genuinely everyone — they sit on every chart and in every order book — so a break of them is a meaningful event with order flow behind it, not an arbitrary line like a daily-anchored VWAP on a 24/7 market. Second, and just as important, the rule barely trades. Prior-day extremes only break a handful of times a month, so the strategy took dozens of trades per symbol across the whole year, not the hundreds or thousands that vaporized the high-frequency rules in this batch on fees alone. Low frequency on meaningful levels is exactly the combination that lets a roughly even win/loss split survive contact with commission — which is why ETH landed at a profit factor near 1.00 and a drawdown in the low double digits rather than the catastrophic curves elsewhere in this research.
Keep the framing honest, though: near breakeven is still not profitable. Both symbols finished marginally in the red, and a strategy sitting on the breakeven line is the one most sensitive to small changes — a worse fee tier, a touch of slippage, or one unlucky cluster of failed breaks flips it negative; a trend filter or a session selection might flip it positive. That sensitivity is the opportunity. When the unfiltered baseline is already this close to even, you are not trying to rescue a broken idea — you are trying to find the modest edge that nudges a near-coin-flip onto the right side of zero. That is a far better starting point than anything else in this batch, and the monthly table shows you which months already did the work.
How to backtest the previous-day break on Secuora
Prior-day high/low levels are a built-in primitive of the AI backtester, so the break rule automates end to end — and because this baseline starts near breakeven, the filter experiments below are unusually worth running.
- Open /backtest/ai and describe it in plain English: "go long on the first hourly close above the previous-day high, short on the first close below the previous-day low, structure stop, 2R target." It compiles to the same deterministic engine that produced this page.
- Run it on BTC and ETH with costs on and confirm you reproduce the near-breakeven result — especially the ETH profit factor around 1.00, the standout of this research.
- Because you start so close to even, the filters matter most here: add a higher-timeframe trend so you only break in its direction, take only breaks during a chosen session, or require the break to clear the level by a buffer — and watch whether expectancy crosses zero.
- Sign up free and use the replay terminal — draw yesterday’s high and low as horizontal levels and replay the next day bar by bar with simulated stop and limit orders to feel which breaks held and which were traps.
- Journal the breaks you would actually take, with confluences and screenshots, so a strategy this close to the line is graded on the exact rules you tested.
Methodology, in one paragraph
Data: Binance spot 1-minute klines (data-api.binance.vision), aggregated per strategy timeframe, June 1, 2025 – June 1, 2026 (12 months). Execution: Secuora’s deterministic strategy runner (the same engine behind the in-app AI backtester) — single position at a time, entries at the close of the signal candle, commission 0.05% per side; no spread/slippage modeled (btc/eth spot spreads are sub-basis-point); position size capped at 10× notional leverage, starting balance $10,000, 1% risk per trade. Swings are confirmed fractals with no look-ahead. These are mechanical results: no discretion, every signal taken. Past performance does not predict future results; this is research, not financial advice.
Frequently asked questions
What is the previous-day high/low break strategy?
You mark the high and low of the prior trading day as fixed levels, then trade the first break of either: long on a close above the previous-day high, short on a close below the previous-day low. Prior-day extremes are among the most-watched levels on any chart, so a break of them is treated as a meaningful intraday event.
What win rate does the previous-day break have?
A sub-35% win rate on hourly BTC and ETH in our test — which is fine for a 2R target, where winners only need to be roughly twice the size of losers. The reason this page leads with profit factor instead is that ETH came in at about 1.00, the closest-to-breakeven result in our entire research series; the exact figures are in the table above.
Does the previous-day high/low break work on crypto?
It came closer than almost anything else we tested. Both BTC and ETH finished marginally negative after costs, but ETH’s profit factor was around 1.00 — gross profit and gross loss nearly equal. It is not profitable raw, but it is the standout near-breakeven baseline in this research, mainly because it trades infrequently on genuinely meaningful levels, so fees never buried it.
How do I backtest the previous-day break myself?
Two ways on Secuora: describe it in plain English at /backtest/ai — prior-day high/low levels are a built-in primitive that compiles to the same deterministic engine used here — or sign up free, draw yesterday’s high and low in the replay terminal, and trade the breaks bar by bar with simulated stop and limit orders. Given how close to even it starts, the filter experiments are well worth your time.