Crypto Signal Winrates: What's Real vs Marketing (The Honest Math Guide)

Walk into any crypto signal group’s landing page in 2026 and you’ll see a number in big bold text.

92% winrate. 95% accuracy. 97% success rate. Never a losing week.

These numbers sell subscriptions. They also, in the overwhelming majority of cases, are not true in any meaningful sense.

This article is going to do something unusual for the crypto signals industry: actual math. We’ll calculate what a real 90%+ winrate would imply statistically, why almost no signal service on Earth can sustain one, and how a legitimate provider running a “modest” 60-70% winrate is often vastly more profitable than a fake 92%.

By the end, you’ll have a framework for evaluating any signal service — including ours — that cuts through marketing and reveals the underlying edge. Or lack of it.

Bring a calculator. Or don’t — we’ll do the math for you.

⚠️ Disclaimer upfront: This article is educational content. None of it constitutes financial advice. Every statistic and example below is illustrative. Crypto trading carries substantial risk of loss, including total loss of capital. No winrate — real or advertised — guarantees future performance.

§1 — What “Winrate” Actually Means (And Why Providers Lie About It)

Let’s define terms.

Winrate is the percentage of trades that close in profit out of all trades taken. If a signal service fires 100 signals in a month and 65 close above entry, the winrate is 65%.

Sounds simple. It’s not.

The problem is that “close in profit” has multiple definitions, and unscrupulous providers exploit all of them.

Definition 1: Hit TP1

A trade is “winning” if it hits the first take-profit target, even if the price later reverses and the trader who scaled out partially at TP1 ends up closing the rest for a loss.

The loophole: a trade can hit TP1 at +0.3% and then reverse to your stop-loss at -1.5%. Marketing counts this as a win. Your account calls it a loss.

Definition 2: Any TP hit

A trade is winning if any of TP1, TP2, or TP3 hit before the stop-loss triggered. Slightly more rigorous, but still abuseable — a signal might brush TP1 for a $0.50 move before reversing hard.

Definition 3: Full TP3 hit

A trade is winning only if the signal’s final target completes. This is honest but dramatic — many legitimate trades don’t hit TP3 even when they’re profitable, because the market reverses at TP2.

Definition 4: Net PnL positive (the correct one)

A trade is winning if a trader following the suggested scaling strategy (e.g., 40% out at TP1, 30% at TP2, 30% at TP3) ends up with a net positive PnL after all exits complete.

This is the only definition that reflects reality. And it’s the definition virtually no marketing page uses.

The marketing trick

When a service advertises “92% winrate,” they almost always mean Definition 1 or Definition 2. Under Definition 4 — the one that matters for your account balance — the same service’s winrate might be 55-60%. Which is fine for a profitable system. But it doesn’t sell subscriptions as well as 92%.

First rule of signal evaluation: always ask which definition the advertised winrate uses. If the provider can’t give a clear answer, the number is meaningless.

§2 — The Statistical Reality: Why 92% Is Almost Certainly Fake

Let’s do some math.

Say a signal provider fires 200 signals in a year. If their “true” winrate (under Definition 4) is 92%, what’s the probability distribution of their results?

Using basic binomial statistics, a system with a true 92% win probability over 200 trades would produce:

Expected winners: 184 out of 200
Standard deviation: ~3.8 trades
95% confidence range: 176-192 winners

For comparison, a system with a “modest” 65% true winrate:

Expected winners: 130 out of 200
Standard deviation: ~6.7 trades
95% confidence range: 117-143 winners

Now here’s the key question: what does the market allow?

Crypto markets, especially at the 4H-to-daily timeframes most signal services trade, exhibit roughly random walk behavior with a drift component. Statistical analysis of BTC, ETH, and SOL price data shows that even the best-timed entries with reasonable stop-loss and target placement don’t exceed ~70-75% winrate over large samples. A few outlier strategies — very tight scalping with micro-targets, for example — can push higher, but at the cost of terrible risk/reward ratios (more on this shortly).

For a service claiming 92% winrate across standard swing and intraday setups on BTC/ETH/SOL, the implication is one of the following:

They’ve found an edge that institutional quant funds with billions in R&D haven’t. Possible. Unlikely.
They’re cherry-picking trades. Counting only wins, hiding losses.
They’re using a loose winrate definition (any TP1 brush = win).
They’re fabricating results entirely. Screenshots without public verification.
They’re running a risk/reward so skewed toward losses that hitting a high winrate is mathematically inevitable. More on this next.

At least one of those is true. Usually more than one.

The Nassim Taleb warning

Taleb’s famous example: a turkey is fed every day for 1,000 days. From the turkey’s perspective, the world is safe and farmers are benevolent. Then Thanksgiving happens.

High-winrate signal systems with bad risk/reward ratios work exactly like the turkey. They win and win and win — until one rare, outsized loss wipes out all the small wins combined. Survivorship bias then ensures you only hear from the providers whose Thanksgiving hasn’t arrived yet.

§3 — Winrate vs Risk/Reward: The Only Equation That Matters

Winrate is only half the equation. The other half is risk/reward ratio, often written R/R or R:R.

Risk/reward = (distance from entry to take-profit) ÷ (distance from entry to stop-loss)

A signal with entry at $100, TP at $104, SL at $98:

Risk = $2
Reward = $4
R/R = 2.0 (commonly written as 1:2)

A signal with entry at $100, TP at $101, SL at $95:

Risk = $5
Reward = $1
R/R = 0.2 (or 1:0.2, terrible)

The profitability equation

A trading system is profitable if:

(Winrate × Average Win) > (Loss Rate × Average Loss)

Rearranged for break-even:

Break-even winrate = 1 / (1 + R/R)

Let’s plug in some numbers:

R/R	Break-even winrate	A 65% winrate is	A 92% winrate is
1:1	50%	Profitable (+15%)	Profitable (+42%)
1:2	33%	Very profitable (+32%)	Extremely profitable (+59%)
1:3	25%	Very profitable (+40%)	Extremely profitable (+67%)
1:0.5	67%	Losing (-2%)	Profitable (+25%)
1:0.3	77%	Losing (-12%)	Profitable (+15%)
1:0.2	83%	Losing (-18%)	Profitable (+9%)

Read this table carefully. It’s the whole point of the article.

A system with a 92% winrate at 1:0.2 R/R is LESS PROFITABLE than a system with a 65% winrate at 1:2 R/R.

Let’s verify with concrete numbers.

Example A — The “92% winrate” trap

100 trades, $100 risk per trade, 1:0.2 R/R:

Winners: 92 × $20 profit = $1,840
Losers: 8 × $100 loss = $800
Net: +$1,040 per 100 trades (+10.4% per 100 trades)

Example B — The “boring 65%” system

100 trades, $100 risk per trade, 1:2 R/R:

Winners: 65 × $200 profit = $13,000
Losers: 35 × $100 loss = $3,500
Net: +$9,500 per 100 trades (+95% per 100 trades)

The “boring” system is 9x more profitable. Because the math of trading isn’t “how often am I right,” it’s “how much do I make when I’m right vs how much do I lose when I’m wrong.”

This is why serious trading firms obsess over expectancy (expected value per trade) rather than winrate. A 30%-winrate system with 1:4 R/R is more profitable than a 70%-winrate system with 1:0.5 R/R.

The advertised winrate is a sales number. The R/R is the technical number.

§4 — How to Reverse-Engineer a Signal Service’s Real Performance

Okay, you’ve seen enough numbers. How do you actually evaluate a signal service?

Step 1 — Gather the raw data

Look for signals that show:

Entry price
Take-profit levels (all of them)
Stop-loss level
Outcome (where the trade actually closed)
Timestamps of entry and exit

Ideally, the service publishes this as a monthly recap. If they don’t, request it. If they refuse, assume the data they’d publish wouldn’t flatter them.

Step 2 — Calculate R/R for every signal

For each signal, measure:

Distance from entry to TP1, TP2, TP3
Distance from entry to SL

If you’re scaling out (say 40% at TP1, 30% at TP2, 30% at TP3), calculate the weighted average reward:

Weighted reward = (0.4 × reward_to_TP1) + (0.3 × reward_to_TP2) + (0.3 × reward_to_TP3)

Divide that by the risk (entry to SL) to get the realized R/R for that signal.

Step 3 — Calculate net PnL under realistic scaling

For winners: assume you scaled out as suggested. Sum up the partial closes.

For losers: assume you closed the entire remaining position at the SL. Full loss.

For partial winners (hit TP1 then reversed to SL): the worst category to count. Traders who don’t move SL to breakeven after TP1 end up with these as net losers. Traders who do move SL to breakeven end up with small wins. Ask the provider which behavior they’re assuming in their published stats.

Step 4 — Compute expectancy

Expectancy = (Winrate × Avg Win) - (Loss Rate × Avg Loss)

If expectancy is positive, the system is mathematically profitable (before fees and slippage).

If expectancy is negative, the system is losing money regardless of how the marketing spins it.

Step 5 — Check for survivorship bias

Ask: does the provider delete or retcon losing signals? Some services quietly edit the Discord channel history after a loss, so old members see a different history than new members.

Defense: take screenshots of the signal feed at regular intervals. If the published monthly recap contradicts your screenshots, you’ve caught the provider lying.

§5 — Red Flags in Reported Winrates

Here’s what specifically should trigger suspicion when you see a winrate claim.

Red flag 1: Above 90% with no transparency

Legitimate services publish their methodology for calculating winrate. If a provider claims 92%+ but can’t explain what counts as a win, run.

Red flag 2: No losing trades in the public channel

Every real trading system loses sometimes. If the public channel only has winners, one of three things is happening:

Losses are being hidden/deleted
Only winners are being published (marketing channel vs real feed)
The sample size is too small to trust

Red flag 3: Weekly winrate variance is too low

Real trading has losing weeks. A provider showing a perfectly consistent 85%+ winrate week after week is either running a tiny sample (e.g., 5 signals a week) or smoothing the numbers.

Red flag 4: No R/R reported alongside winrate

If the provider highlights winrate but buries or omits R/R, they probably have terrible R/R and know it. A service with good R/R leads with it.

Red flag 5: Winrate calculated per asset, not per signal

“Our BTC signals have a 95% winrate!” — but they only posted 8 BTC signals all quarter, and the other 40 signals on altcoins had a 48% winrate. Statistical cherry-picking.

Red flag 6: “We don’t count expired trades”

Some services post signals that never get filled (the market didn’t reach the entry), then exclude these from winrate calculations. Technically defensible, but if “expired” is being used to exclude losing setups that would have been filled a few cents away, it’s cheating.

Red flag 7: Claimed winrate hasn’t changed in 6 months

Real statistics fluctuate. A service claiming exactly “87% winrate” for six consecutive months is either updating the number to match marketing targets or not updating it at all.

§6 — The Honest Math of a Legitimate AI Signal Service

Let’s do a worked example using realistic numbers for a well-run AI signal service.

The service:

Asset coverage: BTC, ETH, SOL
Average 30 signals per month
Typical R/R: 1:2.2 (weighted across scaling levels)
Realized winrate (Definition 4, net PnL positive): 62-68% over rolling 6-month windows

The math over 6 months (180 signals):

Assume 65% winrate, 1:2.2 R/R, $100 risk per trade:

Winners: 117 × $220 = $25,740
Losers: 63 × $100 = $6,300
Gross PnL: +$19,440

Subtract trading costs:

Exchange fees at 0.06% per round trip on $5,000 average position: $3 per trade × 180 = $540
Slippage at 0.1% average: $5 per trade × 180 = $900

Net PnL: ~$18,000 on $100/trade risk.

Now annualize and benchmark:

12-month projection: ~$36,000 net on consistent $100/trade
For a trader with $10,000 account sizing at 1% risk per trade: that’s a 360% theoretical annual return

Reality check: that ~360% number assumes perfect execution, consistent sizing, no drawdown panic, and no missed signals. In practice, real traders execute imperfectly and capture 30-60% of the theoretical edge. So a realistic annualized return from this “boring” 65% service: +100-200% on the account, before tax and psychology.

Compare to the fake “92% winrate” service with 1:0.2 R/R:

12-month net: +$20,800 theoretical
Realistic: +$6,000-12,000 after execution friction
Realistic annualized return: +60-120%

The 65% service beats the 92% service by roughly 2x in real-world returns. And that’s before we account for the likelihood that the 92% provider is misrepresenting their numbers to begin with.

§7 — What Ascendant Traders Actually Runs (Transparency Section)

Since we’re in the honest-math business, here’s our actual data.

Current state as of April 2026:

Tracked winrate across BTC, ETH, SOL: ~65% (Definition 4, net PnL positive)
Target winrate after ongoing optimization: 85%
Average R/R across signals: ~1:2 to 1:2.5
Signal expiry: 4 hours after posting (unfilled signals are dropped, not counted as losses)
Reconciliation: signals are tracked against Blofin’s mark price; VIP members can optionally link their Blofin API to reconcile against real fills

Why we don’t advertise a higher winrate:

Because we’d have to lie. The honest number is 65%. With the 1:2+ R/R we maintain, that’s a profitable system — significantly more profitable than a fake 92% at bad R/R, as the math above shows.

Why we’re targeting 85% anyway:

Not because 85% is the magic number. Because tightening our confluence filters (RSI alignment + MACD confirmation + EMA trend agreement + ADX strength threshold + VWAP fair value + volume confirmation + CoinGecko 24h momentum check) should let us skip a subset of marginal setups. If we remove the weakest 15% of our signals, the remaining 85% should have higher winrate while maintaining R/R. That’s the thesis we’re testing.

What we won’t do:

We won’t quietly redefine “winrate” to hit a marketing target. If the optimization works, the reported number goes up honestly. If it doesn’t, we keep iterating. Either way, you’ll see real data.

§8 — The “Winrate” Question You Should Ask Every Provider

One question separates legitimate providers from sketchy ones.

Ask: “Under your winrate calculation, what happens to a trade that hits TP1 at +0.3% and then reverses to hit stop-loss at -1.5%?”

The four possible answers and what they mean:

“That counts as a win, because TP1 hit.”
- This provider is inflating winrate. Legitimate under lazy definitions, but misleading. They might still be a decent service, but take their advertised winrate with skepticism.
“That counts as a partial win — we assume scaled exit at TP1 and SL moved to breakeven, so net PnL is small positive.”
- This provider is using honest accounting. Their published winrate reflects reality, assuming traders move SL to breakeven after TP1. Reasonable.
“That counts as a loss, because the remaining position hit SL.”
- Most conservative accounting. Their winrate will appear lower than other providers, but for the right reason. These are the most trustworthy stats.
“I don’t know” or “Let me get back to you.”
- The provider hasn’t thought rigorously about their own stats. Ask yourself if you want to subscribe to a service where the methodology is this unclear.

There’s no single “correct” answer. But the clarity of the answer tells you how seriously the provider takes their own numbers.

§9 — The Role of Sample Size

One more statistical concept that every signal subscriber should internalize.

If a provider has posted 50 signals and reports 70% winrate, that’s a small sample. The confidence interval around that number is wide — the true winrate could easily be anywhere from 57% to 82%.

If a provider has posted 500 signals and reports 70% winrate, that’s a much more meaningful number. Confidence interval narrows to ~66-74%.

Rule of thumb: Don’t take winrate claims seriously until the provider has at least 100 published signals with verifiable outcomes. Below that, variance dominates and any winrate is largely luck.

This is also why new signal services often have inflated published winrates. In their first month with 15-20 signals, a lucky streak can produce 85-95% winrate by pure statistical noise. The same service, 6 months later with 150 signals, will regress to their true edge — usually 55-70% for legitimate systems, or worse for bad ones.

If a service is under 3 months old and claims 90%+ winrate: that’s not evidence of skill. That’s evidence of small sample variance. Come back in 6 months and recheck.

§10 — A Quick Framework: Evaluate Any Signal Service in 5 Questions

Print this section and use it whenever you’re considering subscribing to any signal group.

What definition of “winrate” does the provider use? (If unclear → skepticism.)
What’s the average R/R across signals? (Below 1:1.5 → not worth it. 1:2+ → good.)
How many signals in the published dataset? (Under 100 → variance. Over 500 → meaningful.)
Are losing trades equally visible as winning trades? (If not → survivorship bias.)
Can I verify the reported performance against my own observations during a free trial? (If not → walk away.)

A service that scores well on all five probably has a real edge. A service that scores well on three and dodges the other two is probably overstating their results. A service that dodges four or five is selling a story, not a trading system.

§11 — Common Misconceptions to Unlearn

Misconception 1: “Higher winrate = better system”

False. As shown above, a lower winrate with better R/R is mathematically superior. Winrate alone is nearly meaningless.

Misconception 2: “A 95% winrate means 5% of trades lose big”

Not necessarily. A 95% winrate service with 1:0.2 R/R has frequent small wins and occasional larger losses — but the “larger losses” might only be 5-10% of account value, not catastrophic.

The real risk in high-winrate systems is emotional overtrading. Subscribers see win after win and increase position size, thinking the system is bulletproof. When the rare loss comes, they’re sized 3x their normal risk, and the loss is devastating.

Misconception 3: “Professional traders have 90%+ winrates”

Wrong. Most profitable hedge funds and prop traders run 35-55% winrates on directional strategies. Their edge is risk management and R/R, not accuracy. Pattern-day-trading scalpers can push higher, but they need institutional-grade speed and slippage control that retail traders don’t have.

Misconception 4: “If the signal service is wrong, they should offer refunds”

No legitimate service offers refunds for individual losing trades. Losses are a structural part of trading. A service that promises refunds for losses is either charging enough margin to cover losses from fees (which means you’re overpaying) or running a Ponzi-like scheme that will collapse.

Misconception 5: “Free signals are worse than paid signals”

Not always. Some legitimate services publish their best signals in the free tier as a marketing tool. The free feed might be slower (delayed delivery), less frequent (2 signals/day vs 6), but the individual signals are often the same quality. Test free before paying.

Frequently Asked Questions

Is a 60% winrate actually good for a crypto signal service?

Yes, if paired with a risk/reward ratio of 1:2 or better. At 60% winrate with 1:2 R/R, the expected value per trade is +20% of risk amount. Over 100 trades, that’s a +120% return on total risked capital. Very profitable.

Can a real signal service actually hit 80% winrate?

Over short samples (20-50 trades), yes, by luck. Over 500+ trades sustained, almost never. Any sustained claim above 75% should be verified extensively before trusting.

What’s the minimum R/R I should accept in a signal?

1:1.5 is the floor. Below that, you need an unrealistically high winrate to be profitable. 1:2 is standard. 1:3+ is strong.

Generally not. Different services may give contradictory signals on the same asset, leaving you with whipsaw positions. Pick one, trade it for 3-6 months, evaluate, then decide.

How do I track a signal service’s real performance myself?

Keep a spreadsheet. Columns: date, pair, direction, entry, TP1/2/3, SL, outcome (TP hit, SL hit, partial), your actual PnL if you traded it. After 100 entries, you’ll have better data than the provider’s marketing page — because it’s your data.

Why doesn’t every signal service just lie about their winrate?

Some do. Serious long-term services don’t, because the ecosystem has gotten better at verification (third-party tracking sites, screenshot archives, community-driven audits). A service that lies eventually gets caught, and the reputational damage destroys the business. Short-term scams still lie because they plan to exit-scam before reputation catches up.

Final Takeaway

Winrate is the most abused metric in the crypto signal industry. Marketing pages lean on it because it’s simple and sells subscriptions. Actual profitability depends on a dozen other factors — risk/reward, expectancy, sample size, execution quality, fee structure, consistency across market regimes — that are harder to advertise but much more important.

Next time you see a signal service bragging about 92% winrate, ask yourself: what’s the R/R? What’s the sample size? What’s the winrate definition? Who verifies the results?

If the answers are vague, the number is vague. If the answers are clear, even a “boring” 65% might be telling you more than you think.

The signals that make you money are rarely the ones with the flashiest numbers. They’re the ones where the math works when you check it yourself.

Do the math. Skip the hype.

⚠️ Reminder: This article is educational only. It is not financial advice. All numbers and examples are illustrative. Past performance of any trading system does not guarantee future results. Crypto trading carries substantial risk of loss, including total loss of capital.

Want to see transparent winrate tracking in action?

Join the free Ascendant Traders Discord → discord.gg/fcPV99aD9z

Observe our free signals for 30 days, calculate the winrate yourself, and compare it to what we publish. Code ascend29 gets you 40% off the first month of VIP if you decide to upgrade.