The Consistency Rule Specifically Targets Trend-Following: What 10,000 Runs on a Real EA Showed [Opinion]
※ This article is an opinion piece. It is not investment advice. Backtests are tests against historical data and do not guarantee future performance. Firm-specific rules are updated frequently, so always check each company's official site. (Last updated: August 30, 2026)
The starting point: the winners' lineup is oddly skewed
Something has been bugging me while looking over prop payout reports.
The people who make it to a payout skew heavily toward trend-following. And on top of that, the firms they're using tend to be ones with a loose consistency rule, or none at all.
Meanwhile, the group that should be the biggest — people doing casual scalping or day trading — isn't even clearing the challenge, let alone getting to a payout.
So I ran actual EAs for what's commonly called over-trading (nampin/gambling-style), trend-following, and one-sided betting and compared them. The short version: both halves of that observation turned out to be structurally correct.
Method: real EA × real server × roughly 10,000 runs
No probability-model simulation was used at all. Everything here is an actual backtest.
| Item | Detail |
|---|---|
| EA | ELDRA (MT5 version) |
| Instrument | XAUUSD (gold), among others |
| Data | Actual feed data from Fintokei-MT5-Server1 |
| Period | September 1, 2025 – August 28, 2026 (249 trading days) |
| Account | ¥20M (JPY-denominated, 100x leverage) |
| Rules | Fintokei challenge plan: profit target +8% / daily loss cap −5% / max DD −10% / min. 3 trading days |
Evaluation uses a "rolling challenge" approach. Against a year's worth of price movement, we shift the start date one day at a time to run 249 different challenges, and count passes, fails, and expirations. This removes the luck factor of "happened to start at a good time." Multiplying this across 50+ configurations gives roughly 10,000 runs in total.
How we built the judgment logic (get this wrong and everything else is wrong)
Prop firm rules reference different figures depending on the item. This backtest matches that.
| Rule being judged | Figure used |
|---|---|
| Daily loss cap / max DD | Equity (balance + floating P&L) |
| Profit target | Closed P&L |
| Consistency rule's "daily profit" | Closed P&L (booked in full on the day it's closed) |
The third one matters most. You must not pro-rate a floating gain's ups and downs as "that day's profit." Even for a position held for three weeks, the full profit lands on the day it's closed.
We reconstructed equity by rebuilding open positions from the trade history and computing floating P&L minute by minute on 1-minute bars. We confirmed the reconstruction was correct using a setting on the EA side. The ELDRA we used has a feature that "auto-withdraws once the day's loss hits 4.25%," so if the reconstruction is accurate, a 4.25% wall should show up in the distribution of daily drawdowns.
| Configuration | 99th percentile of daily drawdown | Max |
|---|---|---|
| Lot 0.5 | 2.97% | 4.14% |
| Nampin (martingale-style averaging) | 4.50% | 6.93% |
The wall showed up exactly where expected. We also ran a control test with every entry mode turned OFF and confirmed zero trades, ruling out any unintended entries mixed in.
Finding 1: without a consistency rule, trend-following wins outright
First, risk needs to be equalized. Running the same lot size would let two-sided entry methods take twice as many trades and end up with double the risk by default. We adjusted lot size per method so annual equity DD landed in the 15–22% range.
| Type | Setting | Annual P&L | Equity DD | Pass rate (no consistency rule) |
|---|---|---|---|---|
| Trend-following | H1, 50-bar reference | +74.4% | 20.3% | 70.4% |
| Trend-following | Both sides, lot 0.4 | +24.0% | 15.1% | 67.3% |
| Trend-following | H1, 100-bar reference | +85.4% | 21.3% | 65.0% |
| Trend-following | H4 bars | +70.4% | 18.5% | 61.5% |
| One-sided (buy only) | Lot 1.2 | +52.5% | 20.9% | 61.2% |
| One-sided (sell only) | Lot 0.5 | +4.4% | 14.1% | 60.3% |
| Over-trading (10-step nampin) | Lot 0.3 | +21.9% | 16.5% | 13.2% |
| Over-trading (BB contrarian) | Lot 0.2 | −24.2% | 30.0% | 0.0% |
| Over-trading (small-wins-big-loss) | Lot 0.2 | −42.6% | 47.0% | 0.0% |
Trend-following 65–70% > one-sided betting 60–61% > over-trading 0–13%.
The ranking matched the hypothesis exactly.
Over-trading patterns can't be rescued even by cutting risk
This part matters. Both the BB contrarian method and the small-wins-big-loss method (wide stop, small quick takes) still show 30–47% equity DD even at lot 0.2, so they never even qualify for the comparison range (15–22%).
A wide stop loss combined with contrarian entries generates DD as a structural property, so it can't be tuned away with lot size. And since the annual result is negative anyway, it never reaches +8%. Pass rate: 0.0%.
Nampin came in at 13.2%. The breakdown of failures shows 116 expirations — positions stuck holding a floating loss, never reaching a confirmed +8% profit before time runs out.
Finding 2: the consistency rule specifically targets trend-following
This is the core of the article. We apply the consistency rule to the same set of methods. The formula is "largest profitable day's profit ÷ sum of all profitable days," with the full profit booked on the day it's closed.
| Setting | Trades | Max-day ratio on passing runs | None | 30% | 20% | Drop |
|---|---|---|---|---|---|---|
| RR10 (take-profit at 10x stop loss) | 296 | 70% | 30.4% | 5.8% | 0.0% | −30.4pt |
| M15 bars (990 trades) | 990 | 50% | 38.9% | 16.0% | 8.2% | −30.7pt |
| H1, 10-bar reference | 699 | 43% | 52.5% | 39.3% | 28.4% | −24.1pt |
| H1, 50-bar reference | 332 | 30% | 70.4% | 58.8% | 52.9% | −17.5pt |
| H4 bars | 228 | 33% | 61.5% | 58.4% | 56.4% | −5.1pt |
| One-sided (buy only) | 299 | 34% | 60.0% | 52.9% | 50.2% | −9.8pt |
| Nampin | 124 | 18% | 13.2% | 13.2% | 13.2% | 0pt |
The more profit concentrates in a handful of big wins, the more it gets killed.
The purest trend-following setup — RR10, with take-profit set at 10x the stop loss — has a max-day ratio of 70% on the runs that pass. It posts an overwhelming annual result of +171.9%, but its pass rate at a firm with a 20% consistency threshold is 0.0%. Not a single run gets through.
By contrast, nampin sits at an 18% ratio and doesn't move a single point from 13.2%, no matter how tight the threshold gets. It simply never triggers the rule. Because it closes small profits every single day, its max-day ratio is already low from the start.
We tested how far this nampin immunity goes across 12 configurations in a separate article (short version: adding a stop loss raises the pass rate to 41.2%, but it still never catches up to trend-following).
The consistency rule punishes methods that cut losses and let winners run, and lets through methods that never cut losses and stack up small gains instead. This is selection running in the opposite direction from its stated intent.
Finding 3: this explains the original observation
The starting observation was this:
People who reach a payout skew toward trend-following, and they cluster at firms with loose consistency rules
Both halves come from the exact same underlying structure.
- Without a consistency rule, trend-following wins outright (65–70%), so trend-following is what survives and wins
- The consistency rule specifically targets trend-following (the purer it is, the more fatal — 0.0% at RR10), so the winners end up clustering at looser firms
In other words, "the winners look like trend-followers" and "they're at firms with loose consistency rules" aren't two separate phenomena — they're two sides of the same coin.
And this also means the consistency rule isn't functioning as a trader-selection device. It rejects a setup earning +171.9% a year while letting through a nampin setup that earns +21.9% a year while quietly accumulating floating losses.
Finding 4: what actually determines the pass rate is equity drawdown
Across 50 configurations, we correlated each metric against the pass rate.
| Metric | Correlation with pass rate |
|---|---|
| Max equity DD | −0.64 |
| Return ÷ equity DD | +0.52 |
| Number of daily-DD fails | −0.49 |
| RR (risk/reward) | +0.23 |
| Win rate | −0.16 |
| Annual P&L | +0.13 |
Neither annual return nor win rate had much to do with the pass rate. What matters is the swing in equity, and its ratio against return.
That DD has a sweet spot.
| Annual max equity DD | Configurations | Average pass rate |
|---|---|---|
| 0–10% | 6 | 14.1% |
| 10–20% | 9 | 43.0% |
| 20–30% | 4 | 62.4% |
| 30–45% | 11 | 47.4% |
| 45–70% | 5 | 39.3% |
| 70%+ | 9 | 24.8% |
Under-10% DD isn't safe — it just never reaches +8%. This held even across different instruments.
| Instrument | Lot | Equity DD | Pass rate |
|---|---|---|---|
| BTCUSD | 1.0 | 11.9% | 35.1% |
| BTCUSD | 2.0 | 20.0% | 76.1% |
| US30 | 10 | 20.7% | 47.3% |
| US30 | 25 | 36.5% | 35.7% |
This isn't a quirk of gold — it comes directly from the +8%/−10% rule structure itself.
Finding 5: the daily loss cap is the "third wall"
The profit target (+8%) and max DD (−10%) are two walls where "hitting either one ends things." But the daily loss cap (−5%) is a third wall sitting outside those two, and it's far more impactful.
| Lot | Annual P&L | Equity DD | Pass rate | Max-DD fails | Daily-DD fails |
|---|---|---|---|---|---|
| 0.5 | +28.8% | 18.9% | 66.9% | 44 | 0 |
| 1.0 | +57.8% | 43.5% | 44.0% | 62 | 76 |
| 2.0 | +114.5% | 66.8% | 34.6% | 42 | 125 |
| 5.0 | +165.5% | 138.0% | 4.3% | 98 | 147 |
Max-DD fails stay roughly flat, but daily-DD fails explode from 0 to 147. The reason is simple: max DD is a wall judged once during the challenge, but the daily cap resets every morning, so if it takes 20 days, it's judged 20 times. The number of walls is different.
Place the daily guard "inside half the cap"
ELDRA has a feature that auto-withdraws for the day once the day's loss hits a set level. We swept only that value (at lot 2.0).
| Withdraw line | Pass rate | Daily-DD fails |
|---|---|---|
| −0.5% to −2.5% | 49–52% (roughly flat) | 10–19 |
| −3.5% | 35.8% | 110 |
| −4.25% | 34.6% | 125 |
| −4.8% | 31.9% | 135 |
Anywhere from 0.5% to 2.5% is roughly flat; past 3.5%, it's a cliff. You don't need to hit an optimal value — just placing it inside half of the cap gets the benefit. That said, this feature only matters once lot size goes up — at lot 0.5, there were zero daily-DD fails across all 249 configurations.
Limits of this backtest
- This is a single instrument (XAUUSD) focused, one-year, single-EA result. Numbers will change under different market regimes
- Execution is modeled on M1 bars (confirmed the pass-rate difference against real tick data is within 1.5 points)
- The equity reconstruction underestimates the annual max by a median of 2.0 points. Daily behavior was confirmed to match the EA's guard value
- Stop loss/take profit and lot sizes were scaled by us for testing purposes and are not the EA author's recommended settings
- The one-sided betting results depend heavily on gold being in an uptrend during this period (buy-only: +51.4%, 60.0% pass; sell-only: +4.4%, 60.3% pass)
How one-sided betting is treated in the rules
Numerically, one-sided betting isn't bad either, at around 60%, but almost every firm bans it outright in the rules.
| Firm | Treatment in rules | Wording |
|---|---|---|
| FTMO | Banned | "overleveraging, overexposure, one-sided bets, account rolling" |
| FundedNext | Banned | "One-Sided Betting = concentrated directional exposure / multiple trades in the same direction" |
| The5ers | Banned (permanent ban clause) | Names one-sided bets specifically |
| Hantec Trader | Banned | "one-sided bets (bets in a single direction only)" |
| PipFarm | Effectively covered | "gambling trades" |
What's notable is that FTMO, which does not use a consistency rule, explicitly bans one-sided bets. Its terms state:
"Trading that goes against actual market practice, or trading conducted in a manner that we reasonably determine could cause us financial, reputational, or other harm" (FTMO Forbidden Trading Practices)
This is a catch-all clause that lets a firm rule a violation even without it being listed. "No consistency rule = one-sided-betting paradise" isn't quite right — it's simply swapping mechanical judgment for discretionary judgment. A real-world case is documented in our account of receiving a One-Sided Bets warning from The5ers.
The real hurdle isn't the challenge
Industry figures show that only about 7% of challenge takers ever reach even a single payout, and only 1–3% of all participants keep receiving payouts for six months or more.
Starting in 2026, a movement has emerged from within the industry to push back on payout denials based on vague clauses. This is the Zero Payout Denial movement. The concern raised is the structure where conveniently-interpretable wording like "gambling behavior" or "consistency mandate" gets invoked at payout time. FundingPips has made over 171,000 payouts totaling $216M verifiable on-chain.
The real comparison shouldn't be difficulty — it should be verifiability of payout track record.
Practical takeaways
If you're trend-following, check whether a consistency rule exists before anything else
This is the biggest conclusion of this article. Without one, 65–70% pass; at a 20% threshold, that can drop to 0% depending on how far your take-profit runs. FTMO, The5ers, TradingCult, and E8 Markets don't use an explicit consistency rule.
If a consistency rule exists, don't let your take-profit run too far
If your max-day ratio on passing runs exceeds the threshold, you don't pass. RR10 (70% ratio) is 0.0%; around RR3 (30–34% ratio) still passes 52–58% even at a 20% threshold. The greedier your take-profit, the worse off you are — the opposite of ordinary discretionary judgment.
Keep annual equity DD in the 20–30% range
Risk that's too small isn't safe — it just never reaches the target (under-10% DD averages a 14.1% pass rate).
The daily loss cap isn't a "buffer to use up"
Withdraw for the day inside half of the cap. Daily-DD fails drop from 125 to 10–19.
Don't pick the "setup that earns the most"
The correlation between annual return and pass rate was +0.13 — essentially unrelated.
FAQ
Q. Between over-trading, trend-following, and one-sided betting, which one clears the challenge?
Comparing at an equalized risk level (equity DD 15–22%), we got trend-following 65–70% > one-sided betting 60–61% > over-trading 0–13%. Over-trading patterns (contrarian, small-wins-big-loss) still show 30–47% DD even at reduced lot size, so they never even land in the same comparison range.
Q. Does the consistency rule affect trend-following?
Extremely so. The more profit concentrates in a handful of big wins, the more fatal it is — the setup with take-profit at 10x the stop loss had a 70% max-day ratio on passing runs, and a 0.0% pass rate at a 20% threshold. A setup earning +171.9% a year doesn't pass a single run. Nampin, in contrast, sits at an 18% ratio and doesn't budge from 13.2% no matter how tight the threshold is.
Q. Why shouldn't floating profit be pro-rated daily?
Because the consistency rule's "daily profit" is booked in full on the day the position is closed. Treating the ups and downs of a floating gain as daily profit would automatically spread out the profit of any method that holds positions longer, making it look like it never triggers the rule. This backtest originally contained that exact mistake, and fixing it changed the evaluation of trend-following significantly.
Q. How much risk should I take?
The sweet spot was an annual equity DD of 20–30% (average pass rate 62.4%). Below 10%, it drops to an average of 14.1%. The same shape appeared with BTCUSD and US30, so this isn't specific to any one instrument.
Q. Where should I set my daily withdraw line relative to the daily loss cap?
As long as it's inside half of the cap, the exact value doesn't matter much. Anywhere from 0.5% to 2.5% gives a roughly flat 49–52% pass rate; past 3.5%, it falls off a cliff. That said, at lot 0.5 there were zero daily-DD fails, so the guard value didn't affect the outcome at all.
Q. How much should I trust this backtest?
It's a result focused on one instrument, one year, and one EA. We validated the equity reconstruction by confirming that the EA's own daily-withdraw setting (4.25%) shows up exactly as a wall in the distribution of daily drawdowns. We also confirmed zero trades in a control test with every entry mode turned off, ruling out any unintended entries.
Sources
- FTMO – Forbidden Trading Practices (explicit mention of one-sided bets and the catch-all clause)
- FundedNext – Restricted/Prohibited Trading Strategies (definition of One-Sided Betting)
- Babypips – The "Zero Payout Denial" Movement
- QuantVPS – Prop Firm Statistics 2026 / Track360 – Prop Trading Industry Statistics 2026
※ Each firm's rules change frequently. Always check the latest version on the official site before purchasing.
Related articles
Written by
Hosono P | the prop firm strategist
I buy challenges with my own money and record everything through to the payout. Recorded payouts: ¥6.1M in total from Fintokei, Fundora and Funded7, plus $4,776 from The5ers (as of September 2026). Author of the semi-discretionary EA "ELDRA".