Research integrity, out loud. Every strategy is pre-registered and multiple-testing corrected. Almost nothing passes — that’s the point.
As of 2026-06-26T05:12:02Z · Live trading: 已封锁
How we judge an edge: RIaaS: each test is pre-registered and multiple-testing corrected (WF, CSCV-PBO, Holm, permutation null, realistic costs). Almost nothing passes — that is the trust guarantee, and the product. Research only; live stays blocked.
6候选策略已测试
0passed (survived correction)
4test families
Timing / entry signals
none passed
Candidate
p (raw)
p (adj.)
PBO
稳健
Verdict
ob_nohtf
0.950
1.000
0.34
1/8
no edge
mom24
0.497
1.000
0.38
2/5
no edge
ob_htf_t12
0.217
1.000
0.53
4/8
no edge
pat_channel20_rt
0.950
1.000
0.66
1/8
no edge
bb20_2
0.950
1.000
0.53
1/8
no edge
ob_htf_t1
0.217
1.000
0.53
4/8
no edge
regime
none passed
Candidate
p (raw)
p (adj.)
PBO
稳健
Verdict
diversification
none passed
Candidate
p (raw)
p (adj.)
PBO
稳健
Verdict
fundamentals
none passed
Candidate
p (raw)
p (adj.)
PBO
稳健
Verdict
Why we publish our failures
Most trading content shows you winners and hides the search that produced them. We do the opposite. Each candidate
above is pre-registered, then run through walk-forward, CSCV probability-of-backtest-overfitting (PBO),
Holm-corrected p-values, a permutation null, and realistic costs. When you correct for how many things we tried,
almost everything that looked profitable was luck — and we say so (no edge). The one
class that holds up lives on /投资. Live trading stays blocked until something
genuinely survives. This honesty is the product.