Fillable  /  the delay curve  /  measured 10 Aug 2026

2 of 6 still held half their edge five seconds late.

You've been told you lose because you're slow. We measured it: every trade mirrored at 1, 3 and 5 seconds behind the leader, priced at the next real fills in the same pool. What separates an edge you can keep from one you can't isn't your execution — it's whether the edge needed speed at all.

Same market. Same three seconds. Opposite outcomes.

DqDhUiMTUe…
Holds up when you're late
1s3s5s
@1s +70.8%  ·  @3s +84.4%  ·  @5s +77.1%   7 trips
B4PGzR2617…
Loses -6.5 points in two seconds
1s3s5s
@1s +5.3%  ·  @3s +11.8%  ·  @5s +1.4%   10 trips

Both were profitable for the leader. Only one is profitable for you — and there is no way to tell them apart without pricing the same trades at several delays.

The shelf — every wallet that came through positive

What a copier actually kept at a 3-second delay, and the shape of the curve from 1 to 5 seconds. Sample size on every one. None of these are certified: the last shelf we published had two wallets clear the gates, and both failed forward. 0 of 2. That is why nothing here carries a verdict.

DqDh
Too few trips to characterise
not enough closed round trips to call a shape
+84.4%
kept at a 3-second delay
HOLDS AT 5s7 trips
2AqF
The 1, 3 and 5s readings don't line up
at this sample we will not call it a pattern
+44.5%
kept at a 3-second delay
HOLDS AT 5s148 trips
B4PG
Too few trips to characterise
not enough closed round trips to call a shape
+11.8%
kept at a 3-second delay
FADES10 trips
4T3d
The Short Fuse
the edge is gone by five seconds
+2.9%
kept at a 3-second delay
BREAKS BY 5s15 trips
4TZT
The 1, 3 and 5s readings don't line up
at this sample we will not call it a pattern
-4.8%
kept at a 3-second delay
BREAKS BY 5s22 trips
3uof
Too few trips to characterise
not enough closed round trips to call a shape
-14.0%
kept at a 3-second delay
BREAKS BY 5s11 trips

And the danger has a clock

Share of the standard cohort not yet below half its entry price, minute by minute.

0%25%50%75%100%051015minutes after graduationminute 5–6: 635 fall at onceminute 0: 100.0% still above -50%minute 1: 92.6% still above -50%minute 2: 87.4% still above -50%minute 3: 83.8% still above -50%minute 4: 81.0% still above -50%minute 5: 79.0% still above -50%minute 6: 62.7% still above -50%minute 7: 54.8% still above -50%minute 8: 49.6% still above -50%minute 9: 47.4% still above -50%minute 10: 45.9% still above -50%minute 11: 43.9% still above -50%minute 12: 42.3% still above -50%minute 13: 41.1% still above -50%minute 14: 40.2% still above -50%minute 15: 39.1% still above -50%
64.9%of tokens still standing at minute six were still standing at minute fifteen
Across the whole cohort only 44.0% finish above −50%.

Clearing minute six is worth about half again on the odds.

What we won't sell you

49.3% of graduated tokens double inside fifteen minutes. It sounds like the story. It isn't one.

3 seconds
median time to the first 2×
n=1,925 doublers
76.8%
of doubles land in minute 0–1
gone before you see it
-1.8%
what 3 seconds costs you on entry
negative — late buyers pay less · n=23,959
2 of 6
edges still standing at 5 seconds
the ones worth having

The double is the graduation spike itself, taken by whoever is fastest. It never reaches a copier, so we don't count it as one.

Your own number

What does the delay cost you?

Paste your wallet into our bot and we run this same measurement on your own trades — every fill mirrored three seconds late, priced at real neighbour fills. Free, no signup. It will refuse to grade you if your sample is too thin, which is the point.

Check my wallet →