Following a feed with a 48% loss rate, a run of five consecutive losses is more likely than not within 100 trades. The streak is arithmetic. What it costs you is a choice made before it starts.
Why does a streak matter more than a single loss?
Because losses compound in a way that single trades do not, and because the streak is the thing that ends accounts rather than the individual trade that started it.
One losing trade at a sensible size is a rounding error. Six in a row at the same size is a different account, and the follower who sized for the first one usually did not size for the sixth. What makes this worth writing down is that the streak is not bad luck. It is the expected behaviour of any sequence with a loss rate, and the loss rate in published signal feeds is high enough that long runs are ordinary.
Everything below is arithmetic. None of it depends on whether the feed you follow is any good.
How often do calls actually lose?
More often than they win, in our corpus, once every published call is scored under one consistent rule. As of August 2026, across the 1,514 outcomes in our as-published replay that carry a profit-and-loss value:
| As-published replay, outcomes with a P&L value | Count | Share |
|---|---|---|
| Finished in profit | 475 | 31.4% |
| Finished in loss | 728 | 48.1% |
| Finished flat | 311 | 20.5% |
That is one replay run under one rule set, and its figures must not be mixed with our one-take ranking metric, which counts different things on a different denominator. Both rules and what each can establish are on the Methodology page.
The payoff sits close to symmetric: the average winner returns 5.17% and the average loser costs 4.73%, which means a follower needs to be right about 47.8% of the time to break even, and this corpus is right 31.4% of the time. The gap is where the average result of minus 0.65% per outcome comes from. Why a hit rate on its own settles so little is worked through in Win rate: why 90% accuracy can still lose you money.
For everything below, the number that matters is the 48.1% loss rate.
How long a streak should you expect?
Longer than feels reasonable, and long enough that a follower who treats a run of five as evidence of something has misread ordinary arithmetic.
At a 48.1% loss rate, treating each call as independent:
| Sequence length | Median longest losing run | Chance of a run of 5 or more | Chance of a run of 8 or more |
|---|---|---|---|
| 30 calls | 4 | 32.0% | 3.5% |
| 50 calls | 4 | 49.1% | 6.4% |
| 100 calls | 5 | 75.3% | 13.2% |
| 200 calls | 6 | 94.2% | 25.3% |
Read the third row. Follow a feed for a hundred calls, which is a few months on an active channel, and a run of five consecutive losses is more likely to happen than not. A run of eight arrives about one time in eight.
Independence is the generous assumption here, and it is worth naming. Signal calls are correlated: they cluster in the same market conditions, in the same direction, often on the same day. Correlated sequences produce longer runs than independent ones, so the table above is a floor rather than a forecast. When a feed goes cold it tends to go cold on several calls at once, and a follower running three positions from the same channel can lose three at a time.
What does a streak cost at a fixed fraction?
Less than most people fear at small sizes, and far more than most people expect at sizes that feel modest. The arithmetic is multiplicative, so it punishes size disproportionately.
| Loss per trade | After 5 losers | Recovery needed | After 8 losers | Recovery needed |
|---|---|---|---|---|
| 2% of equity | 90.4% remains | +10.6% | 85.1% remains | +17.5% |
| 5% of equity | 77.4% remains | +29.2% | 66.3% remains | +50.7% |
The right-hand columns are the part that surprises people. Losses and gains are not symmetric, because a gain is computed on what is left after the loss.
| Drawdown | Gain needed to get back to even |
|---|---|
| 20% | 25% |
| 50% | 100% |
| 80% | 400% |
| 90% | 900% |
Half your account is not half the problem. It is a requirement to double what remains, from a feed whose average outcome in our corpus is negative.
Where does leverage actually enter this?
Not in the streak, which is a property of the calls, but in the size of each step and in whether you are still holding a position when the market comes back.
Leverage does three things to the sequence above, and none of them changes the loss rate.
- It multiplies each step. A 2% adverse move costs 2% of the notional whatever the leverage. What changes is what fraction of your account that notional represents, and leverage is what lets a small account carry a large notional.
- It moves the liquidation price closer. At 20x, the whole margin behind a position is a 5% move. At 200x, the highest multiple stated in our parsed signals as of August 2026, it is 0.5%. A trade that would have recovered does not get the chance, because the position ended before the recovery.
- It converts a drawdown into a floor. A losing streak at low leverage leaves an account smaller. The same streak at high leverage can leave the account below the minimum size the next call requires, at which point the sequence has ended for reasons unrelated to whether the next call was good.
That second point is the one worth sitting with. A stop is a decision to leave; a liquidation is the exchange making it for you at the worst available moment. What that looks like on a real account is set out in What account blow-ups really mean.
Why do streaks feel like something went wrong?
Because a streak arrives as a story rather than as a statistic, and every story has a cause in it. The channel got worse, the market changed, the last call was obviously bad.
Sometimes that is true. Usually it is a run of five out of a distribution that produces one in roughly three sequences out of every four, when each sequence is a hundred calls long. The trouble is that the two look identical from inside, and the response they invite is the same one: size up to make it back.
That response is what turns an ordinary drawdown into a terminal one. The account is smaller, so the same percentage is a smaller amount, so the temptation is to increase the position exactly when the equity supporting it has fallen. Both halves of the earlier arithmetic point the other way.
The behavioural version of this is well studied in retail trading generally. Our own contribution is narrower and it is the number above: the feed you are following loses 48.1% of the time under a consistent rule, so a run of five is a thing that happens, repeatedly, without anyone doing anything differently.
What actually protects a follower here?
Three things, and only one of them is about the channel.
Size chosen for the sequence rather than the trade. The question to ask of a position is what eight of them in a row would cost, and whether the account still functions afterwards. How to convert a stop distance into a size is in Position sizing when the trade idea is not yours.
A cap on simultaneous positions. Three open calls from one feed are close to one position of triple the size, because they will lose together. And three channels can be one feed: three of the seventeen publishing accounts in our index carry the same signals under different brands, which is visible as 298 outcomes each with hit rates within a fifth of a percentage point. The count of what is actually independent is in Only 14 of the 47 signal channels we index still post.
A rule decided in advance about when to stop following. Written down before a streak, it is a plan. Written down during one, it is a reaction. The only useful version specifies a number of consecutive losses or a drawdown percentage, and specifies it while the account is intact.
Does a channel's advertised win rate change any of this?
Only if it is real, and it is usually computed on a denominator that excludes the calls that hurt. A percentage advertised on a sales page is not the loss rate you will experience.
Three exclusions do most of the work. A call whose entry was overtaken before it filled costs the follower nothing, which is fine, and it also leaves the percentage without a trade that might have lost. In our corpus 1,554 of 3,167 outcomes are cancelled that way as of August 2026, though one channel supplies 1,197 of them and across the other sixteen publishers the rate is 21.9%, as detailed in Half the trades never happen. Then there are the calls with no stop, which never resolve as losses at all, and 51% of parsed signals carry none as of August 2026. Last, a ladder of targets lets the first one hit be reported as a win on a trade that later reversed.
So the streak arithmetic should be run on the loss rate you can verify rather than the one advertised. Building that number yourself takes about ten minutes and the method is in How to check a signal channel's track record in 10 minutes.
What this does not prove
Nothing here says a losing streak means a channel is bad, and nothing says it means the channel is fine. A run of five losses is weak evidence either way, which is exactly the point: it is consistent with a feed that has an edge and with one that does not.
The streak table assumes independent calls, which understates real run lengths because signal calls are correlated. The drawdown table assumes a constant percentage loss per trade, which no real sequence delivers exactly. Both are there to show the shape of the arithmetic, not to predict any particular account.
Our figures describe our index as of August 2026: 1,514 outcomes carrying a P&L value in the as-published replay, from 17 publishing accounts out of 50 indexed. That is not a random sample of the market, the figures move as the replay catches up, and the as-published and one-take rules are never mixed.
Nothing here is financial advice, and nothing here alleges misconduct by any channel. A feed that loses more often than it wins may be doing exactly what it says it does.
The practical read
Three questions, and the useful time to answer them is before the streak rather than during it.
What is the loss rate of the feed I am following, computed by me rather than advertised? What does a run of eight losses at my current size leave me holding? And what have I decided, in advance, to do if that run arrives?
Current ratings for the channels we can score are at Signal Providers.