Guide
How Long Do Losing Streaks Last?
By Agatha Research Desk. Updated
Record numbers refresh hourly; last refreshed .
The math
Every bet is a separate draw. With a win rate p, each bet loses with chance 1 - p, and a run of k straight losses starting at any given bet has chance (1 - p) to the power k. Over hundreds of bets there are hundreds of places a run can start, so a run that looks rare on any one day becomes likely somewhere in a season.
The tables below are exact, not simulated: they count every way a sequence of bets can fall. At a 55% win rate, going from 500 to 1,000 bets adds about 0.9 to the longest run to expect. 52.4% is the break-even win rate at -110.
| Bets | 50% | 52.4% break-even at -110 | 55% | 58% | 60% |
|---|---|---|---|---|---|
| 100 | 6.0 | 5.6 | 5.3 | 4.9 | 4.6 |
| 250 | 7.3 | 6.9 | 6.4 | 5.9 | 5.6 |
| 500 | 8.3 | 7.8 | 7.3 | 6.7 | 6.4 |
| 1,000 | 9.3 | 8.7 | 8.1 | 7.5 | 7.1 |
| 2,500 | 10.6 | 9.9 | 9.3 | 8.6 | 8.1 |
| Bets | 50% | 52.4% break-even at -110 | 55% | 58% | 60% |
|---|---|---|---|---|---|
| 100 | 17.0% | 12.3% | 8.4% | 5.2% | 3.6% |
| 250 | 38.4% | 28.9% | 20.3% | 12.8% | 9.2% |
| 500 | 62.5% | 49.9% | 36.9% | 24.3% | 17.7% |
| 1,000 | 86.1% | 75.1% | 60.4% | 42.9% | 32.4% |
| 2,500 | 99.3% | 96.9% | 90.2% | 75.5% | 62.6% |
Exact counts for independent bets at a fixed win rate.
Try your own numbers
Enter your own number of bets, win rate and typical price. The run figures are exact; the drawdown figures come from 1,000 simulated runs at one unit per bet.
Losing streak and drawdown calculator
Up to 5,000
Your long-run win rate, 1 to 99
American1.909 decimal
The run length you want the chance of
Drawdowns are the deepest fall below a running high, at one unit per bet, from simulated runs with a fixed seed. The run math is exact.
Over 500 bets at a 55.0% win rate, expect a longest losing run of about 7.3; the chance of 8 or more in a row is 36.9%; the median worst drawdown is -16.55u.
Drawdowns
A drawdown is the fall from the highest point a running total has reached to the lowest point after it, counted in units at one unit per bet. It is usually deeper than the longest losing run suggests, because losses do not need to come in a row to add up.
From Agatha's record
Worst drawdown at Agatha's own prices, one unit per pick
To see what luck alone does at Agatha's own prices, the same 3,127 prices were replayed 2,000 times with no edge at all, each pick winning exactly as often as its price implies. The median worst drawdown was -53.25u, and 1 run in 10 fell to -92.38u or deeper. Agatha's actual worst is -131.90u, and 0.8% of the no-edge runs fell at least that far.
Paying the margin changes the picture. A bettor with no skill who pays a 4.55% hold on the same prices has a median worst drawdown of -155.64u, and 70.6% of those runs fall as far as Agatha's worst. The margin, not the streaks, is what sinks a bettor with no edge.
| Bettor | Worst drawdown | Deepest 1 in 10 | Reach the record's worst |
|---|---|---|---|
| Agatha's recordnow -130.88u from its +32.02u high | -131.90u | - | - |
| No edge at all | -53.25u | -92.38u | 0.8% |
| No skill, 4.55% hold | -155.64u | -213.41u | 70.6% |
For the two replays, worst drawdown is the median across 2,000 runs of the same prices.
Agatha's drawdown is counted pick by pick in start order, from a +32.02u high reached during October 13, 2026. The running total chart further down counts at each board day's close, so its high can read lower.
From Agatha's record
Agatha's own streaks
- Longest losing run
- 7 losses
- Longest winning run
- 14 wins
- Current run
- 1 win
- Longest losing run expected by chance
- 7.9
- Board days up / down
- 14 / 28
- Most losing days in a row
- 7
- Losing days in a row now
- 4
Agatha's longest losing run is 7 picks. Her longest winning run is 14. At her loss rate of 38.1% over 3,127 decided picks, chance alone would produce a longest run of about 7.9, and a run of 7 or more 89.6% of the time.
By board day the record has 14 up days and 28 down. The longest run of losing days is 7, and the record is on 4 losing days in a row now. The best day was October 4, 2026 (+21.10u on 99 picks, 73-25-1); the worst was November 3, 2026 (-19.13u on 68 picks, 32-36). These day totals count every graded pick; the calendar on the track record counts only Agatha's Top Picks board, so the same day can read differently there.
Every pick behind these numbers is on the track record, graded and linked to its sealed receipt.
High +25.01u at the close of October 12, 2026. Now -98.87u. Counted at each board day's close, over every graded pick.
Show the numbers by day
| Board day | Picks | Record | Day | Running total |
|---|---|---|---|---|
| October 1, 2026 | 54 | 30-24 | -9.79u | -9.79u |
| October 2, 2026 | 47 | 34-13 | +5.82u | -3.97u |
| October 3, 2026 | 121 | 78-43 | -8.58u | -12.54u |
| October 4, 2026 | 99 | 73-25-1 | +21.10u | +8.55u |
| October 5, 2026 | 57 | 35-22 | -2.61u | +5.94u |
| October 6, 2026 | 49 | 35-14 | +4.17u | +10.12u |
| October 7, 2026 | 73 | 49-24 | +6.40u | +16.51u |
| October 8, 2026 | 57 | 39-18 | +3.60u | +20.11u |
| October 9, 2026 | 65 | 40-25 | -0.56u | +19.55u |
| October 10, 2026 | 125 | 72-52-1 | -10.75u | +8.80u |
| October 11, 2026 | 70 | 46-23-1 | +2.81u | +11.61u |
| October 12, 2026 | 64 | 44-20 | +13.40u | +25.01u |
| October 13, 2026 | 72 | 42-30 | -2.03u | +22.98u |
| October 14, 2026 | 54 | 33-21 | -2.06u | +20.93u |
| October 15, 2026 | 49 | 26-23 | -7.10u | +13.83u |
| October 16, 2026 | 50 | 31-19 | -2.87u | +10.95u |
| October 17, 2026 | 146 | 88-58 | -2.91u | +8.04u |
| October 18, 2026 | 63 | 31-32 | -14.22u | -6.18u |
| October 19, 2026 | 49 | 33-16 | +1.51u | -4.67u |
| October 20, 2026 | 56 | 28-28 | -13.45u | -18.12u |
| October 21, 2026 | 46 | 30-16 | +3.12u | -15.00u |
| October 22, 2026 | 52 | 32-20 | -2.95u | -17.95u |
| October 23, 2026 | 60 | 37-23 | +0.50u | -17.45u |
| October 24, 2026 | 156 | 93-63 | -3.46u | -20.92u |
| October 25, 2026 | 70 | 41-29 | -7.40u | -28.31u |
| October 26, 2026 | 78 | 45-33 | -3.94u | -32.25u |
| October 27, 2026 | 73 | 44-29 | -7.58u | -39.83u |
| October 28, 2026 | 87 | 49-38 | -9.01u | -48.84u |
| October 29, 2026 | 65 | 38-27 | -7.18u | -56.02u |
| October 30, 2026 | 55 | 34-21 | -6.93u | -62.95u |
| October 31, 2026 | 137 | 96-41 | +17.36u | -45.59u |
| November 1, 2026 | 82 | 52-30 | +1.92u | -43.66u |
| November 2, 2026 | 108 | 66-42 | -4.91u | -48.57u |
| November 3, 2026 | 68 | 32-36 | -19.13u | -67.71u |
| November 4, 2026 | 63 | 37-26 | -8.70u | -76.40u |
| November 5, 2026 | 66 | 37-29 | -11.23u | -87.63u |
| November 6, 2026 | 58 | 42-16 | +7.80u | -79.82u |
| November 7, 2026 | 159 | 109-50 | +5.58u | -74.25u |
| November 8, 2026 | 88 | 49-39 | -12.15u | -86.40u |
| November 9, 2026 | 40 | 27-13 | -0.39u | -86.79u |
| November 10, 2026 | 68 | 44-24 | -4.04u | -90.83u |
| November 11, 2026 | 31 | 15-16 | -8.04u | -98.87u |
The honest read
A drawdown as deep as Agatha's worst (-131.90u) happens to a bettor with no edge at these prices in about 0.8% of runs. Luck alone rarely goes that deep, which points to more than a bad run.
The win rate is 61.91% against the 63.78% the prices needed, 1.87 points short. A win rate over 3,127 decided picks is only known to within about 1.74 points either way. The gap is wider than two standard errors, which luck alone rarely produces: the prices have asked for more than the picks delivered.
What to do in a streak
- Keep stakes flat. Every Agatha pick is one unit, win or lose; nothing is sized up after a loss or down after a win.
- Do not chase. Raising stakes to win back a drawdown makes the next losing run cost more, and losing runs keep arriving at the same rate.
- Set limits before you start: a budget for the week and a point where you stop for the day. Decide them while you are calm, not in the middle of a run.
- Take breaks. Time away costs nothing, and the math on this page keeps producing streaks whether you watch or not.
Questions
How long can a losing streak last?
Longer than it feels like it should. At a 52.4% win rate, the break-even at -110, the longest losing run to expect over 1,000 bets is about 8.7, and the chance of 10 or more in a row is 26.8%.
Does a losing streak mean the picks are bad?
Not on its own. Compare the streak with what chance produces at the same win rate and number of bets, then look at closing line value, which settles far sooner than results.
What is Agatha's longest losing streak?
7 picks. At her win rate over 3,127 decided picks, chance alone produces a run at least that long 89.6% of the time.
Should I raise my stakes to win it back?
No. A losing run does not make the next bet more likely to win, and a bigger stake makes the next run cost more. Flat stakes keep every bet the same size, which is how Agatha's record is kept.
How big should each bet be to survive a streak?
Small enough that the longest run you should expect costs a slice of your bankroll, not a chunk of it. The Kelly calculator shows how fast the answer shrinks when your win chance is a little lower than you think.
What is a drawdown?
The fall from a running total's highest point to its lowest point after it. See the glossary.