Why chasing doesn't work
Chasing the draw with a rising stake does not beat the price. Replayed over every no-draw run in this database at the bookmaker's closing odds, a Fibonacci line opened after 5 matches without a draw returns between −14.14% and −2.78% of what it stakes, at every bank size from 3 to 25 bets. The system does not create an edge; it changes the shape of the loss — many small wins, then one large loss that arrives later.
How the line works
Start with one unit on a draw. After each loss, move to the next Fibonacci stake: 1, 1, 2, 3, 5, 8, 13, 21. Stop when a draw arrives or the bank runs out.
This illustration uses decimal odds of 3.460, the median winning price for lines opened after 5 matches without a draw. It holds that price fixed to explain the arithmetic; the historical replay uses each winning match’s own price. Return includes the winning stake; net subtracts every stake in the line. All amounts are in first-stake units.
| Line closes on bet | Total staked | Return | Net |
|---|---|---|---|
| 1 | 1 | 3.46 | +2.46 |
| 2 | 2 | 3.46 | +1.46 |
| 3 | 4 | 6.92 | +2.92 |
| 4 | 7 | 10.38 | +3.38 |
| 5 | 12 | 17.30 | +5.30 |
| 6 | 20 | 27.68 | +7.68 |
| 7 | 33 | 44.98 | +11.98 |
| 8 | 54 | 72.66 | +18.66 |
| 9 | 88 | 117.64 | +29.64 |
| 10 | 143 | 190.30 | +47.30 |
| 11 | 232 | 307.94 | +75.94 |
| 12 | 376 | 498.24 | +122.24 |
| Line still open after 16 bets | F(18) − 1 = 2,583 | 0.00 | −2,583.00 |
What the replay found
47,942 no-draw runs; lines opened at 10, 15 and 20 are later slices of the same runs. N is the number of matches without a draw before the line opens; its first bet is on the next match.
| Start run N | Lines replayed | Median bets | p90 bets | p99 bets | Longest ever | ROI across banks |
|---|---|---|---|---|---|---|
| 5 | 47,942 | 3 | 8 | 16 | 46 |
−14.14% (79 priced busts) to −2.78% (25 priced busts)
Every bank: ROI and priced busts
3 bets:
−5.17% (11,263 priced busts)
4 bets:
−5.08% (8,306 priced busts)
5 bets:
−4.37% (6,118 priced busts)
6 bets:
−4.75% (4,548 priced busts)
7 bets:
−4.31% (3,335 priced busts)
8 bets:
−4.21% (2,466 priced busts)
9 bets:
−5.60% (1,874 priced busts)
10 bets:
−5.91% (1,402 priced busts)
11 bets:
−6.40% (1,063 priced busts)
12 bets:
−7.44% (819 priced busts)
13 bets:
−7.49% (622 priced busts)
14 bets:
−6.37% (466 priced busts)
15 bets:
−6.11% (347 priced busts)
16 bets:
−7.52% (270 priced busts)
17 bets:
−9.14% (212 priced busts)
18 bets:
−9.56% (163 priced busts)
19 bets:
−8.19% (122 priced busts)
20 bets:
−9.71% (96 priced busts)
21 bets:
−14.14% (79 priced busts)
22 bets:
−7.40% (55 priced busts)
23 bets:
−11.79% (46 priced busts)
24 bets:
−7.97% (36 priced busts)
25 bets:
−2.78% (25 priced busts)
|
| 10 | 10,178 | 3 | 9 | 17 | 41 |
−14.41% (79 priced busts) to +7.26% (15 priced busts)
Every bank: ROI and priced busts
3 bets:
−4.32% (2,466 priced busts)
4 bets:
−6.34% (1,874 priced busts)
5 bets:
−6.58% (1,402 priced busts)
6 bets:
−7.03% (1,063 priced busts)
7 bets:
−8.15% (819 priced busts)
8 bets:
−8.05% (622 priced busts)
9 bets:
−6.66% (466 priced busts)
10 bets:
−6.32% (347 priced busts)
11 bets:
−7.80% (270 priced busts)
12 bets:
−9.48% (212 priced busts)
13 bets:
−9.84% (163 priced busts)
14 bets:
−8.35% (122 priced busts)
15 bets:
−9.90% (96 priced busts)
16 bets:
−14.41% (79 priced busts)
17 bets:
−7.47% (55 priced busts)
18 bets:
−11.92% (46 priced busts)
19 bets:
−8.03% (36 priced busts)
20 bets:
−2.77% (25 priced busts)
21 bets:
+7.26% (15 priced busts)
22 bets:
−2.46% (13 priced busts)
23 bets:
too few busts to judge (9)
24 bets:
too few busts to judge (7)
25 bets:
too few busts to judge (5)
|
| 15 | 2,362 | 3 | 9 | 17 | 36 |
−14.96% (79 priced busts) to +7.56% (15 priced busts)
Every bank: ROI and priced busts
3 bets:
−9.02% (622 priced busts)
4 bets:
−6.72% (466 priced busts)
5 bets:
−6.24% (347 priced busts)
6 bets:
−8.11% (270 priced busts)
7 bets:
−10.03% (212 priced busts)
8 bets:
−10.33% (163 priced busts)
9 bets:
−8.56% (122 priced busts)
10 bets:
−10.20% (96 priced busts)
11 bets:
−14.96% (79 priced busts)
12 bets:
−7.52% (55 priced busts)
13 bets:
−12.16% (46 priced busts)
14 bets:
−8.08% (36 priced busts)
15 bets:
−2.66% (25 priced busts)
16 bets:
+7.56% (15 priced busts)
17 bets:
−2.39% (13 priced busts)
18 bets:
too few busts to judge (9)
19 bets:
too few busts to judge (7)
20 bets:
too few busts to judge (5)
21 bets:
too few busts to judge (4)
22 bets:
too few busts to judge (4)
23 bets:
too few busts to judge (3)
24 bets:
too few busts to judge (3)
25 bets:
too few busts to judge (2)
|
| 20 | 577 | 3 | 10 | 16 | 31 |
−17.15% (79 priced busts) to +8.39% (15 priced busts)
Every bank: ROI and priced busts
3 bets:
−13.35% (163 priced busts)
4 bets:
−9.87% (122 priced busts)
5 bets:
−11.68% (96 priced busts)
6 bets:
−17.15% (79 priced busts)
7 bets:
−7.84% (55 priced busts)
8 bets:
−13.02% (46 priced busts)
9 bets:
−8.32% (36 priced busts)
10 bets:
−2.42% (25 priced busts)
11 bets:
+8.39% (15 priced busts)
12 bets:
−2.21% (13 priced busts)
13 bets:
too few busts to judge (9)
14 bets:
too few busts to judge (7)
15 bets:
too few busts to judge (5)
16 bets:
too few busts to judge (4)
17 bets:
too few busts to judge (4)
18 bets:
too few busts to judge (3)
19 bets:
too few busts to judge (3)
20 bets:
too few busts to judge (2)
21 bets:
too few busts to judge (2)
22 bets:
too few busts to judge (2)
23 bets:
too few busts to judge (1)
24 bets:
too few busts to judge (1)
25 bets:
too few busts to judge (1)
|
Start runs 10–20 are later slices of the start-5 runs; at large banks a swing of two or three busts changes the sign in every row, so a positive figure there is not evidence of an edge.
Follow a starting run: sign in to replay it with your own bank. Median, p90 and p99 count bets until the draw, not odds. ROI is profit divided by total money staked across priced lines, shown as a range over banks covering at least 3 bets meeting the priced-bust display floor; it is not flat-stake ROI or profit divided by the bank.
The shape of the loss
Why a bigger bank does not help
A bigger bank lets the line survive more losing bets, but a bust then loses more units. For start run 5, every bank shown still has negative ROI: postponing the loss has not created an edge.
Start run 5; smallest, middle and largest indexed banks, in first-stake units. For an even number of banks, the middle is the lower of the two central entries.
| Bank units | Bets covered | Lines that bust | ROI |
|---|---|---|---|
| 4 | 3 | 39.29% | −5.17% |
| 986 | 14 | 1.62% | −6.37% |
| 196,417 | 25 | 0.10% | −2.78% |
Bust shares use all 47,942 closed lines; ROI uses the 28,109 lines whose winning bet had a price. Lines still open at the end of the archive and lines censored by coverage gaps are excluded from both.
What this does not claim
This is an in-sample description of the historical archive, not an out-of-sample test or a promise about future returns. Only ninety-minute draws are measured. Coverage varies by competition and era; the index tables do not store the first and last match dates.
The indexer uses Pinnacle closing draw prices, with the first available Bet365 draw price as fallback; that fallback can be an opening price. Its flat-stake calibration uses Pinnacle closing prices only. Bookmaker names and the fallback rule come from the indexer: the summary tables do not store bookmaker labels or counts by source.
Computed 2026-09-30 by the draw-market indexer over 47,942 no-draw runs; lines opened at 10, 15 and 20 are later slices of the same runs