What the paper test teaches
162 closed trades (123 to learn from, 39 in the last 14 days to check on). Updated 2026-11-22 22:42 UTC.
Rules tried in shadow
Each rule is applied afterwards to the same trades; the real picks don't change. A rule is proposed only when it gains R both in the learning period and in the last 14 days, which it was not chosen from. It is adopted only by a person.
| Rule | Skips (or changes) | Gain R (all) | Learning | Last 14 days | Status |
|---|---|---|---|---|---|
| Long only when BTC is above its daily cloud, short only when it is below Ichimoku's own rule: trade with the bigger trend. It keeps working when the market turns, unlike "only shorts". | 37 | +1.22 | +2.11 | -0.89 | shadow (too few trades yet) |
| Only setups where all 6 indicators agree Fewer, stronger setups: a 5-of-6 setup has one indicator against it. | 82 | -8.23 | +5.33 | -13.56 | shadow |
| Skip setups whose stop is closer than one ATR (the coin's normal 4h move) to the entry A stop inside the normal move of one 4h candle is hit by noise, not by the trade going wrong: half the stops in the test were hit within hours. | 122 | -15.94 | +8.29 | -24.23 | shadow |
| Skip setups whose stop is 2% or more from the entry A wide stop on a 4h setup often means a choppy coin; a stop that far is hit before the move starts. | 48 | +13.2 | +23.02 | -9.82 | shadow |
| Enter only when the 4h Tenkan crossed the Kijun in the trade's direction at most 3 candles ago The moment matters: a fresh cross is the start of a move, an old one may be near its end, and with no cross at all the short-term trend is against the trade. | 114 | +8.6 | +25.62 | -17.02 | shadow |
| Skip setups where the price is 2 ATR or more from the 4h Kijun A price far from the Kijun is stretched and tends to snap back to it before going further. | 74 | +5.18 | +11.94 | -6.76 | shadow |
| Close early when the 4h Tenkan crosses the Kijun against the trade Get out when the short-term trend turns, instead of waiting for the stop. | 6 | +3.23 | +3.23 | 0 | shadow (too few trades yet) |
| Only trades with the coin's own daily trend: daily Tenkan above the Kijun for a long, below for a short The bigger timeframe first: a 4h setup against the coin's daily trend swims against the current. | 28 | -2.2 | +1.64 | -3.84 | shadow (too few trades yet) |
| Enter only when the 1h Tenkan crossed the Kijun in the trade's direction at most 6 candles (hours) ago A finer moment than the 4h chart: wait until the hourly chart turns the trade's way too, and not long after. | 113 | -4.51 | +14.17 | -18.68 | shadow |
| Close early when the 1h Tenkan crosses the Kijun against the trade Reacts sooner than the 4h cross, at the cost of more false alarms. | 32 | -21.12 | -11.78 | -9.35 | shadow |
No rule proposed: none gains in both periods with enough trades yet.
By direction
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| long | 47 | 21% | -0.36 | -16.99 |
| short | 115 | 47% | +0.47 | +53.84 |
By market phase (BTC against its daily cloud) and direction
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| long, BTC above cloud | 38 | 16% | -0.67 | -25.54 |
| long, BTC below cloud | 3 | 33% | -0.12 | -0.35 |
| long, BTC in cloud | 6 | 50% | +1.48 | +8.9 |
| short, BTC above cloud | 34 | 35% | -0.03 | -0.87 |
| short, BTC below cloud | 45 | 67% | +1.32 | +59.19 |
| short, BTC in cloud | 36 | 33% | -0.12 | -4.48 |
By how many of the 6 indicators agreed
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| 4 of 6 | 14 | 29% | +0.33 | +4.58 |
| 5 of 6 | 68 | 35% | +0.05 | +3.65 |
| 6 of 6 | 80 | 45% | +0.36 | +28.62 |
By stop distance
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| 2-4% | 45 | 27% | -0.25 | -11.27 |
| 4% or more | 3 | 33% | -0.64 | -1.93 |
| under 2% | 114 | 45% | +0.44 | +50.05 |
By stop distance in ATR (the coin's normal 4h move)
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| 0.75-1 ATR | 91 | 40% | +0.21 | +19.44 |
| 1-1.5 ATR | 25 | 48% | +0.29 | +7.31 |
| 1.5 ATR or more | 8 | 25% | -0.4 | -3.21 |
| under 0.75 ATR | 31 | 29% | -0.11 | -3.5 |
| unknown | 7 | 71% | +2.4 | +16.81 |
By how fresh the 4h Tenkan/Kijun cross was
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| 0-1 candles ago | 21 | 57% | +0.8 | +16.73 |
| 2-3 candles ago | 12 | 33% | +0.01 | +0.12 |
| 4-10 candles ago | 33 | 30% | -0.13 | -4.32 |
| more than 10 candles ago | 54 | 41% | +0.17 | +9.09 |
| no cross in the trade's direction | 27 | 22% | -0.5 | -13.37 |
| unknown | 15 | 67% | +1.91 | +28.6 |
By distance from the 4h Kijun in ATR
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| 0.5-1 ATR | 19 | 21% | -0.47 | -8.95 |
| 1-2 ATR | 46 | 46% | +0.46 | +20.98 |
| 2 ATR or more | 74 | 35% | -0.07 | -5.18 |
| under 0.5 ATR | 8 | 38% | +0.17 | +1.4 |
| unknown | 15 | 67% | +1.91 | +28.6 |
By how fresh the 1h Tenkan/Kijun cross was
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| 0-2 hours ago | 7 | 43% | +0.17 | +1.18 |
| 3-6 hours ago | 12 | 50% | +0.41 | +4.9 |
| 7-24 hours ago | 48 | 38% | +0.12 | +5.7 |
| more than a day ago | 23 | 48% | +0.37 | +8.41 |
| no cross in the trade's direction | 42 | 26% | -0.23 | -9.6 |
| unknown | 30 | 50% | +0.88 | +26.26 |
By the coin's own daily trend (daily Tenkan against Kijun)
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| against the daily trend | 28 | 39% | +0.08 | +2.2 |
| unknown or flat | 18 | 61% | +1.58 | +28.39 |
| with the daily trend | 116 | 36% | +0.05 | +6.26 |
By exchange
| Trades | Win rate | Avg R | Sum R | |
|---|---|---|---|---|
| Binance.US | 125 | 40% | +0.27 | +33.7 |
| Gate | 5 | 0% | -1.2 | -5.99 |
| Kraken | 31 | 45% | +0.33 | +10.22 |
| MEXC | 1 | 0% | -1.08 | -1.08 |
BTC-USDT against its Ichimoku cloud on the daily chart (Binance.US), from the daily candles closed before the pick: no look-ahead. Timing: on the 4h candles of the trade's own exchange, closed before the pick: how many candles ago the Tenkan crossed the Kijun in the trade's direction (none: not crossed that way), and how far the entry was from the Kijun in ATR; the same cross on 1h candles; and whether the coin's daily Tenkan was on the trade's side of its Kijun. Entry rules are judged on the real R; the early-exit rules on a replay of each trade on 4h or 1h candles with and without the exit (R before fees, so only the difference counts). Kraken serves only its latest 720 hourly candles, so older Kraken trades have no 1h facts. Past results say nothing certain about the future; this is a paper test, not advice.