Index Cadence TechnoTrader

Index Cadence is a rules-based intraday system, packaged as an EA. It trades one instrument: the US tech index CFD (USTEC, also sold as NAS100, US100 or NDX100 depending on your broker - the EA finds the name itself). Around the US cash open it measures how far this market normally travels by each time of day, using the last 14 sessions, and draws a band from that. A 15-minute candle that closes outside the band is the signal: above the band it buys, below it sells. Every position carries a hard stop from the moment it opens, and everything is closed before the end of the trading day. It does not hold overnight and it does not hold over the weekend. One chart, one instrument, and roughly 275 trades a year in the test window described below.

This is not a holy grail and we will not sell it as one. It is one measured edge, executed with discipline, on real historical data. The test window starts in 2020 because that is where our index CFD history starts - we did not choose it. It happens to open inside the COVID crash of March 2020 and to include the 2022 rate shock, but it is a short window and you should read it as one. Every figure on this page is either a measurement or an inference we label as one. None of it is a promise about your account.

The system at a glance

Before you run this on any account, read this page to the end and look through the screenshots. That is not a formality. Two things you will need are only shown properly in the images: the table of how often losing streaks occur, and the comparison of average win against average loss. MetaQuotes does not translate screenshots, so those pictures are the one part of this listing that every buyer in every language reads identically, and we put the least comfortable numbers there on purpose. The sections below on losing streaks, on drawdown and on what this test cannot tell you are the ones worth your time - the performance figures are the easy part.

How to judge this system

Judge it in months and quarters, never trade by trade. The same test, one number per horizon:

Our recommendation to every buyer, without exception: run the free demo in your own strategy tester first, on your own broker's data. Then run it on a demo account and watch how it behaves - how it enters, what a normal losing week looks like - and only then move to a live account, starting small.

And once it runs, let it run. This system behaves the way the numbers above describe only if nobody interferes with it: no closing a position by hand because it looks wrong an hour in, no skipping a day that feels risky, and above all no switching it off after a bad week. Every protection a position needs is already inside - the hard stop, the daily and weekly breakers, the close before the end of the day. The stretch you would most want to switch it off is usually the stretch that comes before the recovery, and turning it off there is how a positive system produces a negative account.

It suits someone who is not at a screen all day. Check it once a day or once a week, and judge it in months. Living beside a system that keeps its own rules is, in itself, practice for keeping yours.

The EA closes its positions before the end of the trading day, so by design it does not carry overnight or weekend gap exposure. As with any EA, that depends on your terminal being connected when the flat time arrives. That is the main practical difference from a swing system: you normally start the next morning from a known position, which is none.

What the backtest shows (2020.01 - 2026.06, M15, leverage 1:100)

Three risk packages, measured on a 3000 USD account. Each level was measured separately, not scaled from one another:

Those three drawdown figures are the worst that happened on one path. Out of sample the same mechanism drew down roughly twice as deep, so plan the Balanced package for 25 to 35 percent, not 17. The section headed The honest part explains why, and it is the section we would read first.

Account size at the default Balanced package:

Below about 2000 USD the broker's minimum lot starts refusing trades, and it does not refuse them at random: the ones it drops are the widest-stop trades, which are the trades of the most volatile weeks. At 1000 USD roughly one signal in a hundred is lost that way. At 500 USD it is closer to one in eight, and that is exactly why the 500 USD line shows the highest profit factor on this page. A higher profit factor from a narrower sample is not a better system, it is a smaller one. We are publishing the 500 USD row because it is true and because you would find it yourself, not because we recommend starting there.

The same test was repeated on real tick data instead of 1-minute bars: 56.0% per year across the same window, drawdown 16.8%, profit factor 1.35, with an identical number of trades. Intraday systems usually lose a lot to tick-level modelling; this one loses under one percentage point, because its entries are on closed 15-minute candles and its stops are wide relative to the spread.

Why a 42% win rate is the wrong thing to look at

Fewer than half of these trades make money, and that is by design rather than in spite of it. Over the test window the average winning trade made 261 and the average loser lost 142, a payoff of 1.84 to 1. In risk terms the average win was 1.59 times the amount risked and the average loss was 0.88 of it, which leaves an expected 0.175 of one risk unit per trade. That small edge, repeated 275 times a year, is the whole product.

The most telling pair of numbers is the extremes: the largest single winner in the test made 3919 and the largest single loser lost 567. That asymmetry is structural, not luck. A hard stop caps what a loss can become; nothing caps a winner except the end of the trading day.

We are pointing this out because a high win rate is the easiest statistic in this market to manufacture and the most commonly used to sell. Let losers run, close winners early, and 85% of your trades close in profit while the account bleeds. Many EAs achieve exactly that with a wide or absent stop, where a single loss erases four, six or more wins. Ours is the other shape, and it is the shape you can check yourself: win rate, average win against average loss, and profit factor are all in the report the free demo generates on your own data. Judge any system by the last two, never by the first.

Losing streaks

Read this section before you buy. With a 42% win rate, runs of losses are arithmetic, not malfunction. In the test window:

If you buy this and see four losses in your first fortnight, nothing has gone wrong. That is a normal week here, and it will happen to you about eleven times a year. The equity curve in the screenshots is built out of exactly these stretches.

And the important caveat: 11 is the longest streak that happened on one path through one history. It is not a ceiling. A longer run than anything in this test is entirely possible in live trading, and the account you run this on has to be able to sit through one.

The honest part

The 17% drawdown is a floor, not an expectation. It is the worst drawdown that happened on one particular path through one particular six and a half years, and that single realised number carries a large sampling error. When we measured this same mechanism on an out-of-sample window, its drawdown came out about twice as large as in-sample. Apply that honestly and the realistic expectation for the Balanced package is a drawdown in the 25 to 35 percent range, not 17. Plan your account for the larger number.

The out-of-sample evidence is thin and we would rather say so than hide it. The index CFD history on our data source starts in 2020, so the only pre-2020 test we could run was on the S and P 500 index for 2017-2019. It produced a profit factor of 1.07 - positive, but weak. Worse, when we split it: one year carries all of it (2018), while 2017 and 2019 were roughly flat to slightly negative, and the longest run of losing months out of sample was four, against two in sample. The dominant trade direction also flipped between the two windows. That is a real warning sign and it belongs on this page, not in a drawer.

One instrument, one bet

This system trades one instrument. That is deliberate - we measured adding a second US index and a European index, and both diluted the result rather than diversifying it, because these markets move together intraday. But it means you are buying one bet, not a portfolio. There is no second engine to carry a bad stretch on the first.

Two thirds of the profit in the test comes from the first 45 minutes after the US cash open. If your broker's spread on this instrument widens sharply at that moment, your result will differ from ours by more than the usual amount. Check it before you rely on any number here.

What no test on this page can show

Two features cannot be judged from any test on this page. The news filter is invisible to the strategy tester, so its effect on your results is genuinely unknown until you run it live. And prop mode changes the risk level, which changes everything downstream of it - the numbers above describe the normal setting, not that one.

Finally, what a test can and cannot tell you. This one covers six and a half years that include the COVID crash of March 2020, the 2022 rate shock and the tightening cycle after it, and the out-of-sample window adds the fourth quarter of 2018. So it is not a quiet sample, and we did not pick a calm stretch on purpose. But six and a half years is a short life for a trading system, and it cannot contain everything a market can do: a regime this instrument has not met, a policy shift, a conflict, a change in how index CFDs are priced or traded, or simply a stretch where this particular edge stops paying. A historical simulation shows how a fixed set of rules behaved against prices that already happened. It cannot tell you what your account will do, and past results do not guarantee future ones. Read every figure on this page as a measurement of the past, and size your risk for the possibility that the future turns out worse than it.

What it does not do

• No martingale
• No grid
• No averaging down
• No hidden recovery tricks
• No holding a loser overnight to avoid booking it

Every position carries a hard stop from the moment it opens. The EA never caps your profit and never stops trading because you are "up enough".

Optional prop-firm rule protection

Prop accounts are a natural fit for this system for one structural reason: it is flat by the end of every day, so it never carries the overnight exposure that most firms restrict. A separate input block adds rule protection on top of that: a static drawdown tracker measured from your starting balance, a daily loss tracker, and an automatic flatten well before the firm's limit rather than at it.

Turning prop mode on also forces the risk down to 0.40% per trade, and it does so deliberately. We simulated a 10 percent target with a 10 percent static drawdown limit and a 5 percent daily limit, starting a challenge on every single trading day of an out-of-sample window. At our normal 1.00% setting, 25.3% of those attempts breached a rule. At 0.40%, 0.7% did. In-sample data does not show this difference at all - it reports roughly the same outcome at every risk level, because a challenge ends at the target before the deep drawdown ever arrives. That is why the risk cap is not a suggestion in this mode.

What prop mode costs: in the same test window, 17.4% per year with a worst drawdown of 6.8%, against 57.2% and 16.9% in the normal setting. Roughly a third of the return, for a drawdown that fits inside a 10 percent limit.

This block protects the rules. It does not promise you will pass a challenge and we will not pretend otherwise. Passing depends on the market during your attempt, and our own measurement says roughly one attempt in three was still running rather than finished when the test data ran out.

One practical warning for prop accounts, which we ran into ourselves. We tried to measure this system on a prop firm's own server and could not: the firm's index history was a single day long. Prop platforms often carry very little price history, because their business is the challenge and not the archive. That matters here because everywhere else on this page we tell you to verify our numbers on your own broker's data before trusting them - and on a prop account you may simply not be able to. If that is your situation, run the demo on a normal broker with the same instrument first, and treat the prop account as execution rather than as a place to test.

News filter and trade randomisation

Both exist because prop firms ask for them, not because they improve results.

The news filter pauses new entries around high-impact releases in the instrument's own currency. There is one thing you must know before you rely on it: the economic calendar is not available inside the MetaTrader strategy tester. We verified this - the call returns error 4014. That means the news filter cannot be backtested by us or by you, and every figure on this page was measured with it switched off. Turn it on and the EA will take fewer trades than any test shows. How many fewer, nobody can know in advance. The EA prints a warning into the journal whenever it runs in the tester with the filter on, rather than silently pretending to filter.

Randomisation makes your entries slightly different from other accounts running the same system, which is what firms look for when they check for copied trading. It shifts the lot size by up to a few percent in either direction, which shifts your risk by the same few percent. We measured its cost on two different random seeds: one run came out 1.5% below the unrandomised result and the other 1.1% above it, with the drawdown and profit factor unchanged. In other words the cost is inside the noise, and it goes both ways. Set a fixed seed if you want a run you can repeat.

Broker data and costs

Read this section before you buy. The headline figures on this page come from one primary data source: a raw-spread server that charges no separate commission on this index CFD (IC Trading / Capital Point). If your broker charges a round-trip commission on index CFDs, enter it in the commission input so the EA's own cost gate can see it - otherwise the gate will let through trades whose stop is too small to be worth taking on your account.

We have since run the same configuration on four more brokers. Their index histories start on different dates, so every comparison below was made on the window both feeds actually cover, never on the date range printed at the top of a tester report - that header will happily show six years while the data underneath it starts three years in.

So across five brokers the result runs from 9.9% worse to 5.9% better, with the profit factor between 1.34 and 1.42 and the win rate between 42.3% and 44.2%. None of the five charges a separate commission on this instrument - the cost sits entirely in the spread, which we checked deal by deal rather than assuming. Worth noting for a practical reason: those five brokers call this instrument three different things, USTEC, NAS100 and NDX. The EA resolves the name itself, so you do not have to know which one yours uses.

That is a real result and we would rather you weigh it than take our word for it, so here is what it does not cover. Five brokers is five, not the market. A sixth could sit outside that range, and a broker whose index spread widens sharply at the US open - which is where two thirds of this system's profit is made - would hurt more than any of these did. We are still not going to recommend a broker we have not measured. Run the free demo in your own strategy tester on your own broker's data before you rely on any number here; that result, not this page, is what your account would have seen. Broker names are mentioned as test data sources only; we are not affiliated with any of them.

About the price

This price reflects what we can currently prove, and no more than that: five brokers' data over one market era, no live track record on this engine yet, and out-of-sample evidence we have described above as thin. As that evidence grows - a longer out-of-sample record, months of live behaviour to point at - the price will follow it upward. We would rather tell you that plainly than have you wonder later why it moved.

Session timing

The whole mechanism is anchored to the US cash open, so the EA has to know when that is on your server's clock. Out of the box it is set for servers on GMT+2 or GMT+3, which covers most European brokers, and it corrects itself on the weeks when the US and European clock changes disagree. If your server sits on a different offset, set the server GMT input once and the EA recomputes the whole session from the New York open. At startup it prints the session anchor, the entry cut-off and the flat time into the journal - check that line once against when your index market actually opens.

There is also a pause around the year turn. It is ON by default, and both dates are yours to move, or you can switch it off entirely. Liquidity between the holidays is thin, and in our test skipping that stretch improved the result slightly at an unchanged drawdown.

Requirements

• Hedging account (the EA refuses netting accounts loudly at start)
• Leverage 1:100 or higher recommended; every published figure was measured at 1:100
• Broker must offer a US tech index CFD. The EA resolves the common broker names automatically and a suffix input covers the rest. A missing symbol is reported loudly, never skipped silently.
• Works from 1000 USD (measured above). Attach it to any one chart you like - the symbol and timeframe of that chart are both ignored. The EA trades the instrument you name in the settings, on 15-minute bars, which is the only timeframe it was measured on and the only one it uses. If it cannot find your broker's name for the index, it refuses to start and says so; it will never quietly trade the chart symbol instead.
• Strategy tester: 1 minute OHLC is a valid model for this EA and real ticks give nearly the same result. "Open prices only" is not supported.
• Every broker other than our test source is UNTESTED. Run the free demo in your own strategy tester first. "It gives different numbers at my broker" is expected behaviour, not a defect.

Support: ask in the product's Comments tab; honest questions get honest answers. If you have run the demo on a broker we have not tested, we are genuinely grateful when you share the result in the Comments - every shared test makes the picture more honest for the next buyer.