Does a positioning portfolio of three long and three short rules beat holding BTC from 2025?
Yes, modestly, and mostly thanks to one rule. Six rules, three that buy BTC and three that sell it, made +6.6% from January 2025 to October 2026 with a worst fall of 13 percentage points of capital. Holding BTC ended at -1.5% with a worst fall of 69 points. The portfolio was positive in both 2025 and 2026, but one long rule made most of the gain, and the rate is far below the roughly +45% a year the same rules made from 2020 to 2024.
The rules were picked the way PekkaLabs's LB6 system was built: LB6 is six simple rules ("engines"), three long and three short, found by trying many combinations of options-market measures and keeping the best.
How it differs from the previous try
The previous report, "Does a portfolio of positioning rules, built the LB6 way, beat holding BTC from 2025?", picked six rules from the same search. All six were long, and all six needed one thing: few trader accounts long. One idea in six forms; it went flat from 2025.
This portfolio is built from the same 29,160 rules (two measures, a cut-off for each, long or short, held 24 hours after the rule was last true, fees and funding included, scored on September 2020 to December 2024), with three changes, written down before it was picked:
- the 3 best long and 3 best short rules, as LB6 has three long and three short engines;
- each measure may appear in at most 2 of the 6 rules;
- each rule's daily results must not move too closely with those already chosen (correlation below 0.5).
Rules are ranked by their efficiency (return divided by worst fall), taken as the middle value over the rule and the up to eight rules with each cut-off one step higher, lower or the same, so that a lone lucky cut-off cannot win. That ranking score was 11.8, 9.0 and 6.5 for the long rules below and 5.9, 3.5 and 2.0 for the short ones; the table shows each rule's own efficiency. Each rule gets a sixth of the capital.
The six rules
| rule | 2020-24 trades | 2020-24 efficiency | 2025-26 trades | 2025-26 result |
|---|---|---|---|---|
| long when few accounts are long (ratio at most median) and buyers dominated the last 24 hours (top 10%) | 155 | 15.8 | 96 | -2.6% |
| long when few accounts are long (at most 60th percentile) and BTC has not risen much in 24 hours | 254 | 9.9 | 100 | -10.1% |
| long when funding rose over 24 hours and open interest is low against its weekly average (bottom 30%) | 291 | 6.5 | 110 | +31.2% |
| short when open interest is high against its weekly average (top 10%) and the long/short ratio rose (top 30%) | 98 | 7.6 | 13 | +14.3% |
| short when open interest grew over 24 hours and the long/short ratio jumped (top 10%) | 160 | 3.3 | 55 | +8.1% |
| short when funding jumped (top 10%) and sellers dominated the last 24 hours (bottom 20%) | 74 | 2.5 | 6 | -1.1% |
The long/short ratio is the number of trader accounts long for each account short. Open interest is the total of open futures positions. Funding is what longs pay shorts every 8 hours. Percentiles are of each measure's values over 2020 to 2024.
Two of the short rules read as one idea: sell when new positions pile in and more accounts turn long, a crowd arriving late. The third (funding jumps while sellers dominate) is a different one and barely traded from 2025. The two long rules carried over from the previous portfolio lost from 2025; the new long rule and two of the short rules made the gain.
Results

| portfolio | holding BTC | |
|---|---|---|
| return 2020 to 2024 | +197% | +292% |
| worst fall, 2020 to 2024 | 14 points | 126 points |
| efficiency, 2020 to 2024 | 14.5 | 2.3 |
| return Jan 2025 to Oct 2026 | +6.6% | -1.5% |
| worst fall, 2025-26 | 13 points | 69 points |
| efficiency, 2025-26 | 0.51 | -0.02 |
Returns are hourly moves added up, not compounded; "points" are percentage points of capital.
| year | 2020 (Sep-Dec) | 2021 | 2022 | 2023 | 2024 | 2025 | 2026 (to Oct) |
|---|---|---|---|---|---|---|---|
| portfolio | +34% | +68% | +22% | +26% | +47% | +3.0% | +3.6% |
| holding BTC | +95% | +84% | -83% | +103% | +93% | +3.3% | -4.8% |
How much to trust it
- The search makes portfolios look good by itself. Running the whole process on scrambled data (each measure shifted in time by a month or more against BTC's price, so it no longer lines up with it) gave portfolios with an efficiency of 8.4 in a typical run, and one run reached 19.0, above the real 14.5. The 2020-2024 numbers are therefore mostly the search; the 2025-2026 result is the part that counts.
- 2025-2026 was not blind. It had been used by the two earlier searches, and the yearly results of their long rules there had been seen. The portfolio rule (3 long, 3 short, at most 2 rules per measure) was chosen after seeing that the all-long portfolio failed there, though not after seeing these six rules.
- Some rules lean on a few trades. The third short rule got 84% of its 2020-2024 result from its 5 best trades and traded only 6 times from 2025; the first two short rules about half. The long rules are more spread out (20% to 36%).
- One rule carries 2025-2026. The funding-and-open-interest long rule made +31.2% on its own share; without it the other five together come to about +1.4%. The best short rule's +14.3% comes from 13 trades.
- In 2025 alone the portfolio (+3.0%) made a little less than holding BTC (+3.3%), with a far smaller fall; the gain over holding comes from 2026.
- The edge shrank. About +45% a year before 2025, about +3.6% a year after.
What it means
This is the first candidate strategy from the positioning data: a mix of buying when funding rises while open interest is low, and selling when positions and long accounts pile in. It has made a little money, with a worst fall about twice its gain, through a period when BTC went nowhere with a fall five times deeper.
The way to know more is the third step agreed for these searches (search, then a check on later data, then a forward run): freeze the six rules as they are and run them forward on data that does not exist yet. Binance's archive adds each day's data the next day, so this needs no live trading, only a monthly re-run. At the rate these rules traded in 2025-2026 (about 380 trades over 22 months across the six), 30 trades take about two months.
Limits
- One rule shape (two measures, 24-hour hold, no stop), one exchange, BTC only.
- Ten scrambled runs give a rough sense of how good the search makes things look.
- Returns are summed at 1x, without compounding or position sizing.