Today, the 14-day ATR of S&P 500 futures is more than twice its usual level (its median over the past six months).
On days like that, price can jump past the stop, and the loss can be far bigger than planned.
A jump like that is rare, but it's possible.
So on those days, I size each trade so that a loss twice as big as planned still doesn't cost me my account.
Even if risking less than on a normal day costs me some profit when the trade wins, the smaller size lowers the chance of losing my account.
And that matters even more when you're trading with a live account (not a prop firm one), because there the money you lose is your own.
Do you lower your risk on days like today, or keep your usual size?
2
1
203
A piece of advice you'll hear often is to risk more per trade in an evaluation, so you get through it sooner.
A bigger risk does make the evaluation shorter, because you reach the profit target or the drawdown limit in fewer trades.
But fewer trades also means the result depends more on the order your wins and losses happen to come in, and less on what your strategy really makes.
That's why, in my simulations, as the risk per trade went up, the share of evaluations that reached the profit target got closer and closer between a strategy with a small edge and trades with no edge at all.
It's worth simulating your evaluation twice, once with your trades and once with trades that make nothing on average, because the difference between the two is the part your strategy is really adding.
249
Exiting at 4:10 pm instead of at the 4:00 pm New York close made one of my setups earn about 60% more per trade on average.
I had built that setup around the end of the New York session, so exiting at the close felt like the obvious choice.
But price doesn't always stop moving right at the close.
So before you decide when to leave the trades of a setup like that, try a few different exit times around the close, or you may get out while the move is still going.
1
2
371
Does a quiet, narrow range mean a big move is coming?
It's a common belief that when price goes sideways in a narrow range, a big move is coming, or that price bounces off the top and bottom of the range while it holds.
I tested both beliefs on seven years of data, in four different ways and with more than a hundred versions of the rules.
In my newsletter, I show the four ways I tested quiet ranges and the result that surprised me most: tradingamok.substack.com/p/wโฆ
4
488
My daily goal isnโt based on a % gain, but on following my rules 100%.
Thatโs the part I want to get right every day, so I donโt chase results, and a red day, week or even month doesnโt bother me.
I only set result goals one year ahead, and thatโs when I check how close I got.
My daily goal was 1% a day for YEARS
I'm still a huge believer that 1% a day is a very healthy goal for beginner/intermediate day traders
Most beginners blow up by chasing 10% days, 100% days, etc. 1% a day can be achieved with a single +10% trade using 10% of your trading account. That's realistic. That can be doable using a smart strategy & proper risk management
Obviously every day is not going to be +1%. Some days will be smaller green days. Some days will be days with no action. The other goal is to make sure all red days are -1% or smaller
It's a realistic goal that can be achieved using proper techniques, unlike the traders with goals of 10, 25, 100, etc % a day
2
6
520
Nvidia reports its quarterly results on Tuesday, November 17, after the US stock market closes at 4:00 pm New York time.
Nvidia is one of the biggest companies in the Nasdaq, so news about it can move the whole index.
However, the economic calendar usually doesnโt show company news like a results report.
So if you trade Nasdaq futures, itโs worth checking the days when big companies like Nvidia report their results, and not only the economic calendar.
1
1
398
When I compare prop firms, I check how far the account can fall, the consistency rule and the payout rules.
But none of these tell me if the firm will still exist in a few months.
So I'd rather take payouts often, because money in a funded account is only mine once I withdraw it.
5
3
515
โฐ ๐๐ฎ๐๐ ๐ฑ๐ฎ๐ ๐๐ผ ๐ท๐ผ๐ถ๐ป ๐บ๐ ๐ฐ๐ผ๐๐ฟ๐๐ฒ
My course on building and backtesting a strategy with AI started on Thursday, and you can join until 23:59 (CET) tonight.
You can catch up on the first two lessons before tomorrow, when we start with the data.
Then you follow a real setup of mine from the first idea to the final verdict, one lesson every weekday until December 23, with no coding needed.
tradingamok.gumroad.com/l/frโฆ
1
404
On Monday, I had a trade open and wanted to close it before the target. On other days, I've wanted to set a target where my rules have none, or close at breakeven.
Each of those feels like protecting the trade, or even improving it. But the result I expect from each setup comes from a backtest that managed every trade only as its rules say. A setup's rules can include a step like moving the stop to breakeven once the trade is 1.5R in profit, or say to do nothing after the entry.
Some trades in those backtests went against the entry first and then ended in profit. Others kept going past the point where I would have closed them. If I close early, some of the first group can end at zero or as a loss, and the others as smaller wins, so the backtests' average result per trade no longer matches what I actually get.
That's why every setup I use has the same rule after the entry: I manage the trade exactly as its backtest did, and if that means never moving the stop and never closing early, I leave it alone.
Since I started backtesting my setups the way I do now, following my rules has been much easier. I think it's because I see no reason to change something that worked over years of data exactly as it's written.
1
2
540
If you compare two groups of trading days by their average move, a few very big days can fool you.
In the picture, one 100-point day makes the first group's average 20, against 10 in the second.
So also sort each group's days from the smallest move to the biggest, and look at the day in the middle.
Here it's 10 in both, so a normal day moves the same.
353
Don't plan with the number your backtest gives you.
Test many slightly different versions of your rules, like entering a few minutes earlier or later, or using a slightly wider stop.
Then sort them from worst to best, and plan with the one in the middle.
For example, my backtest says one of my setups makes 0.21R per trade, but I plan with 0.13R, which is what the middle one made out of 1,152 versions of the same rules.
If you decide how much to risk with the middle and your setup does better, nothing breaks, but if you decide it with your backtest and it does worse, you've been risking too much the whole time.
2
401
As a discretionary trader, I didn't have years of data behind my setups.
My setups still come from reading price, but now I write each setup as fixed rules and test it on YEARS of data, so before I trade it, I know how often it won and what a win paid and a loss cost on average.
351
While going through the first results of a new setup I'm testing, I noticed that the long trades were doing clearly better than the short ones, and I caught myself thinking that maybe this setup should only take longs.
It felt like a discovery, but it was a guess I'd made while looking at results, so I had it checked like any other setup.
The test took seven years of 1-minute data, split every trade into longs and shorts, and asked two things:
1. Do the shorts lose money? They didn't, because they made less than the longs but they still made some.
2. Could the longs beat the shorts by that much just by luck? To find out, the computer shuffled the "long" and "short" labels among the trades at random many times, and a difference as big as mine showed up fairly often.
So I kept both sides, because dropping the shorts would have meant throwing away trades that earn money just because the longs had looked better in one table of results.
In my course, you learn to check this kind of guess with the data, on a real setup we follow from start to finish.
It started today, and you can join until Sunday: tradingamok.gumroad.com/l/frโฆ
1
324
Before testing a strategy, I put the most recent years of data aside without looking at them, and I open them only once, as the final test.
When a strategy fails there, there's almost always one slice of those years that still made money, like one hour of the day.
It's tempting to keep that slice as the new version, but once I use those years to choose the rules, they can no longer judge them.
And when you check enough slices, one of them will look good just by luck.
So when the final test says no, I close the setup.
If I still believed in it, the only fair way back would be a new test on data it has never seen, even if that takes years.
In my course, you learn what to lock away before the first backtest and how to know when a strategy is really dead, on a real setup we follow from start to finish.
It starts tomorrow, November 12: tradingamok.gumroad.com/l/frโฆ
My first backtests didn't make money, and I thought I'd never find a setup good enough to trade
I felt the same frustration as this trader when I started backtesting my setups with AI.
A backtest goes through years of past prices and records every trade your rules would take.
I tested many setups built on things I was sure were true about price, and the backtests didn't confirm what I believed.
But I kept testing, and several setups have now passed the checks I use to decide what's good enough to trade.
And the checks I have today are harder to pass than the ones I had on my first day, because I've added some I hadn't thought of at the start.
In my experience, most setups fail at least one of these checks, so I never trade them with real money. Some don't pass the first checks. Others fail one of the last ones, like a backtest on recent years that I don't look at until the end, so that test uses prices the setups have never seen.
By then, I've already spent a lot of time on those setups and feel sure they'll pass. That's why a failure at the end is harder to accept than an early one. It's normal to feel disappointed, and I'm even tempted to change something in the setup, so that it can still pass the check it failed.
But a failed check is as useful to me as a passed one, because it shows me a setup that would have made me lose money if I'd traded it.
So the best I can do is be honest with the process, with the results and with myself.
I started building strategies again in November after spending about four months working on my backtesting engine.
Iโve tested four ideas so far and all four have failed my robustness checks. The engine is doing its job by filtering out potentially overfitted strategies but itโs still frustrating spending days researching and developing an idea, only for it to fail validation.
Mixed emotions. One half of me is proud that the engine is doing exactly what I built it to do, while the other half is disappointed that I havenโt managed to develop a single strategy that passes my checks so far.
2
2
770
The bond market is closed today, so I checked whether I should still trade.
It's Veterans Day in the US, a holiday for the bond market but not for the stock exchange, which is open as usual.
It's easy to assume that any holiday means a quiet, strange market, and that it's safer to stay out.
But a holiday for bonds alone is different from a day when the stock exchange doesn't open or closes early. So I wanted to know whether these days behave like a holiday or like a normal day.
I took every day from 2016 to 2025 when bonds were closed and stocks were open, and compared the volume of the Nasdaq and S&P 500 futures I trade with normal days, before the New York open and during the New York session.
In all four cases, volume was what normal days usually show, even in the quietest one, the S&P 500 futures during the New York session.
So today, I trade as on any other Wednesday.
With AI, finding the dates and comparing them with normal days took me a few minutes. That's what I use it for, to check a belief before I let it change how I trade.
In my course, you learn to test holidays and news days as filters, on a real setup we follow from start to finish. It starts on November 12: tradingamoklab.substack.com/
Limit entry or market entry: which one works better?
A common tip is to enter with a limit order on a pullback, because you get a better price and a smaller stop.
I tested it on one of my setups, comparing four ways of waiting for that pullback with simply entering at market once the signal is confirmed.
The better price was real, but my limit orders missed my best trades.
In my newsletter, I show which trades they were and why they got away: tradingamok.substack.com/p/wโฆ
1
2
405
๐ ๐ ๐ฐ๐ผ๐๐ฟ๐๐ฒ ๐๐๐ฎ๐ฟ๐๐ ๐ถ๐ป ๐๐๐ผ ๐ฑ๐ฎ๐๐
My course on building and backtesting a strategy with AI starts on Thursday, November 12.
Until December 23, you follow a real setup of mine from the first idea to the final verdict, and you rebuild each step with Claude.
It starts from zero, with creating your Claude account, and you don't need to know how to code, because Claude writes the code and you learn how to check it.
There are still places, at $99 plus VAT: tradingamok.gumroad.com/l/frโฆ
1
426
A few months ago, I probably wouldn't have taken this trade.
That's because the market had already rallied a lot, so going long that close to the day's top would have felt like buying too high.
But in my test on years of data, this setup's trades made money on average after costs.
Not every one of them wins, but together they do, so I take them without doubting the entry.
In my course, you learn to run that kind of test yourself, on a real setup we follow from start to finish.
It starts on November 12: tradingamok.gumroad.com/l/frโฆ
1
9
743
๐ง๐ต๐ถ๐ ๐ถ๐ ๐๐ต๐ฎ๐ ๐ผ๐ป๐ฒ ๐น๐ฒ๐๐๐ผ๐ป ๐ผ๐ณ ๐บ๐ ๐ฐ๐ผ๐๐ฟ๐๐ฒ ๐น๐ผ๐ผ๐ธ๐ ๐น๐ถ๐ธ๐ฒ
In my course, you build and backtest a strategy with Claude from zero, and each lesson takes between 45 and 70 minutes.
These are three pages from one of them, and they all have the same parts, so you always know where you are:
1. What you'll do in it, and what you need from the earlier ones.
2. For each part, what you're going to do and why, and then a block called "Do this now" with the message you paste into Claude.
3. Where there are numbers, a table with mine, so you can check that yours match.
4. A checklist at the end, so you know you're ready for the next one.
So, when Claude gives you something different, you see it straight away, and the page tells you what to do next.
And if you're still stuck, you ask me in the questions chat.
You do all of this on a real setup of mine that we follow from start to finish, and you don't need to know how to code.
If it isn't for you, you have 14 days to ask for a refund.
It's for a group of 20 people, at $99 plus VAT, and it starts on Thursday, November 12: tradingamok.gumroad.com/l/frโฆ
5
570














