I'm Dan. LA-based futures trader, founder of Stacked Edge, building Blackwell Capital.
What you'll get here:
- Prop-firm rules, broken down plainly
- Risk, sizing and execution: the stuff that decides whether an account survives
- Honest questions traders actually argue about
Follow if you trade futures or funded accounts.
1
1
150
Two green days in a row, and Blackwell has nearly erased Monday’s $832.90 loss. 📈
Blackwell Capital | November 18
🟢 Today’s Net P&L: +$338.05
Two trades. One winner, one loser.
🥇 Williams %R | Short 5 MGC
🟢 +$700.40 net
📉 Morning Flush | Long 5 MES
🔴 −$362.35 net
That brings us to:
📊 Week-to-date: −$28.30
💰 Last two days: +$804.60 net
We started Monday deep in the red. Now we’re less than $30 away from getting the week back into positive territory.
But something else stood out today.
Two gold strategies wanted to trade at nearly the same time.
Williams %R already owned the position, so Blackwell blocked Keltner ADX from entering. Its later exit was also rejected because it didn’t own the trade.
That’s the kind of discipline we need when running 10 automated strategies on one account.
Execution timing:
• Williams entry: 7.82s to local acceptance
• Morning Flush entry: 1.26s to local acceptance
• Both exits confirmed closed locally in roughly 1.3s
We’re still working through Windows clock synchronization, so these aren’t verified broker-fill latencies.
Our market regime detector also continues progressing:
📊 58 usable signal annotations across 9 strategy labels, monitoring trend, chop, and volatility across MNQ, MES, and MGC.
Still shadow-only. No changes to live trading or proven performance improvement yet.
We’re still in overall drawdown, but the last two sessions have put us right back within reach of a green week.
And we didn’t need to increase risk or force extra trades to get there.
From −$832.90 on Monday to just −$28.30 on the week.
Not every recovery happens in one big trade.
Sometimes it’s just about letting the systems do their job, respecting the risk, and stacking good decisions.
Building Blackwell in public. 🏗️
$MGC $MES $MNQ
18
My AI discretionary trading bot took zero trades in NQ today, and that was the bot's call, not a glitch.
The old challenge has been stopped, but the learning keeps going with real paper trades every day. There's no P&L target. I'm just watching how the bot does over time on P&L and win/loss.
What the market did (MNQ, PT):
• It sold off from the 6:30 open to 26,951.75 by 6:50
• My bot re-graded the day from trend-down to range at 6:56 when the 1m gaps flipped bullish
• A second flush to 26,921.5 held later in the morning
• It rallied to a 27,322 double top at 9:30 against the 27,325.5 high
• It faded after the 11:00 FOMC minutes and chopped around VWAP (~27,120–27,150) all afternoon
Why my bot sat out:
• 6:45–6:50: the open was trending down, but every 5m bar made a new low with no base. The bot's new extension filter blocks chasing a move like that, so no short.
• 7:35: the planned short trigger closed below 27,125 but only graded B (thin stop, two-way chop). The bot needs B+.
• 9:35: a short off the double top was a B: no rejection close, and there were unflipped bull gaps just below.
• 12:07 VWAP-reclaim long and 12:28 failed-breakdown long: both B.
Day P&L: $0
Running total since the desk started on Sep 28: +$2,000.50 over 26 trades (early trades were counted as shadow fills while order placement was being fixed)
• Win rate 53.8%
• Profit factor 1.42
• Current drawdown −$380
• Max drawdown −$1,335.50
Honest lesson: the cleanest move of the day, the opening sell-off, came before any base formed, so there was no entry within the rules. On a wide two-way range day, sitting out is a real result. Forcing a B setup is how the bot gave money back earlier in this experiment.
#NQ #futures #daytrading #AItrading
31
Minutes hit. NQ dropped 119 points in about 7 minutes.
Nearly every Fed official now sees more hikes as likely.
On one mini that candle was $2,380. On a micro, $238.
If your daily loss limit is $1,000, one mini on the wrong side of that print is more than two days gone.
The minutes were scheduled. The size was a choice.
So what did you do: go flat into the release, cut size, or trade right through it?
2
1
27
The Nvidia trade is already over. Most people were asleep for it.
NQ spiked to 27,609.25 on the print overnight, then gave most of it back before the bell.
So the gap up on your screen isn't fresh news. It's a gap that's already been sold once.
That changes the open:
The overnight high is a level, not a target.
Fed minutes hit at 11 AM PT, so the morning is a 4.5-hour window.
Size like the news is spent, because it is.
Do you treat a faded overnight high as resistance today, or as the magnet?
58
Flat into Nvidia and watched NQ rip about 1.3% after hours?
You didn't miss a trade. You skipped a coin flip.
That's roughly 350 points.
1 mini = ~$7,000
1 micro = ~$700
Same move the other way on a $2,000 trailing drawdown ends most accounts before the open.
The real trade starts tomorrow at 9:30 ET.
Chase the gap, fade it back toward today's close, or wait 30 minutes and let it pick a side?
1
1
28
My AI discretionary trading bot makes its own calls on NQ futures (paper MNQ). The original experiment is stopped, but the learning keeps going with real paper trades every day. No P&L goal, just tracking how the bot performs over time.
Tue Nov 17: +$262.50 on 2 trades (1W / 1L).
What the market did: NQ broke down early. The bot saw the break under 27,194.5 on a down-trend day, with ES and RTY weak and VIX up. NQ then flushed to an RTH low of 26,659 by 07:50, and the 07:55 5m bar ripped 255 pts back up to close at 26,922.5.
Trade 1, short 3 MNQ: price was already ~20 pts past the break, so the bot did not chase. It put a resting sell limit at 27,205 at 07:22 (pullback into the 1m gap, under the 5m EMA9). Filled 27,208.25. Target 27,113.5 filled ~07:30. +$568.50 (+1.64R).
Trade 2, short 2 MNQ: after the flush, the bot re-priced its resting short to 26,815 (retest of the broken 26,822.5 shelf). It filled near 26,818.5 and got stopped near 26,895 in that same 07:55 reversal bar. -$306.00 (about -0.96R). The 1m bear gaps then flipped bullish, so the bot re-graded the day to range and stayed flat for the rest of the session.
Combined record since Sep 28: +$2,000.50 over 26 trades, 53.8% win rate, profit factor 1.42, current drawdown -$380.00, max drawdown -$1,335.50.
Honest lesson: the first short worked because the bot waited for the pullback. The second came after a very stretched move. The bot's own notes flagged the extension and it took the trade anyway. After a flush like that, a second with-trend short needs more confirmation.
#NQ #futures #daytrading #AItrading
1
68
Yesterday: −$832.90. Today: +$466.55. 📈
Blackwell bounced back with a perfect trading day, going 2-for-2 after yesterday’s rough start to the week.
Blackwell Capital | November 17
🟢 Net P&L: +$466.55
🎯 Win Rate: 100% (2/2)
💰 Gross: +$478.75
💸 Fees: $12.20
Today’s trades:
🟢 EMA Trend | Short 5 MNQ: +$366.40
🟢 Weak/Strong Close | Long 5 MES: +$100.15
Still in drawdown, but today recovered more than half of yesterday’s losses.
The best part wasn’t just the P&L.
Four alerts received. Four processed. Zero rejected entries or failed exits.
And no unnecessary trades trying to force yesterday’s losses back.
Execution timing remains encouraging:
• EMA entry accepted locally in 8.24s, exit confirmed closed in 1.19s
• MES entry accepted locally in 6.97s, exit confirmed closed in 1.13s
Most of the entry time was spent in the system’s signal collection and selection process. We’re still resolving Windows clock synchronization, so these aren’t verified broker-fill latencies.
On the development side, our market regime detector has now reached 52 usable signal annotations across 9 strategy labels.
We’re tracking trend, chop, and volatility across multiple timeframes to better understand which market environments favor each strategy.
Still shadow-only. No proven performance improvement or changes to live execution.
Two green trades don’t erase the drawdown, just like yesterday’s losses didn’t erase the work we’ve put into Blackwell.
We’re still down $366.35 on the week, but the system is running, our execution safeguards are holding up, and we’re continuing to gather data.
A green day with no overtrading and no recorded execution failures? I’ll take that every time.
Building Blackwell in public. 🏗️
$MNQ $MES $NQ $ES
31
Micros aren't for beginners. On days like today they're the only size where a real stop fits.
NQ flushed about 656 points in the 10 AM hour. That's $13,120 on one mini.
A stop that respects that kind of tape is maybe 60 points:
1 mini = $1,200
1 micro = $120
On a $1,500 daily limit, the mini gets one try. 3 micros get four.
Same idea, same stop. The size decides whether you're still trading at noon.
Did you drop to micros today, or keep minis and tighten the stop?
20
Nvidia reports after the cash close. NQ futures will still be open when it hits.
Before you hold anything into it, do this math:
NQ is ~27,400. A 1% move is ~274 points.
That's $5,480 on one mini. $548 on one micro.
Yesterday's range alone was ~1,045 points.
Take your room: daily loss left or distance to your trailing floor, whichever is smaller.
$2,000 of room survives a 1% gap against you on 3 micros. Not 4.
A stop doesn't protect you from a gap. The size is the stop.
Also check your firm's flat-by time. Some close you out before the print anyway.
Save this for every earnings night.
Holding anything through the print tonight, or flat by the close?
29
Made big money shorting today? Check your consistency rule before you request anything.
NQ went from ~28,279 to ~27,235 since Sunday night. Days like that create one huge green day, and that's exactly what a consistency rule is built to catch.
The math with a 30% rule:
$3,000 best day → you need $10,000 total profit before payout.
With a 40% rule → $7,500.
The big day doesn't get you paid faster. It raises the bar for every day after it.
Save this before the next trend day.
Do you keep trading normal size to dilute it, or slow down and grind small green days? Can we have a discussion?
26
Fewer than 1,000 traders affected, and they still announced it publicly before anyone found out at payout time.
That should be the bar for every prop firm. Which firm surprised you at payout instead?
This week, we’ll be restricting access in a handful of countries. This will impact less than 1,000 current Topstep Traders but we want to be transparent with you all about why.
Protecting the integrity of our program and the security of our community is our responsibility. Lately we’ve seen a rise in coordinated activity from certain regions, including attempts to bypass ID verification and payment safeguards, and coordinated trading across accounts.
We’ve been shutting down the accounts involved, but the activity keeps returning through new profiles and new tactics. Topstep exists to help traders grow and prepare for live markets, and we’ve avoided country-level restrictions wherever possible.
Given the recent surge in fraud rings, we’re taking a harder step. Watch our socials for more details about which countries will be impacted and next steps.
51
My AI discretionary trading bot took one trade on NQ today (Mon Nov 16). It lost, and the day finished at -$365.50.
Quick context: the old challenge experiment is over, but the learning continues. The bot still makes its own calls and takes real paper trades every day. There's no profit target and no loss number it has to hit. I'm just tracking how it does over time, in P&L and in wins and losses.
What the market did (times PT):
• The 6:30 open flushed through the overnight low (~27,612), swept down to 27,485, then ripped back up to 27,717. It whipsawed around VWAP with the 5m ATR near 2x normal.
• By 8:00 it had printed a 27,338.25 low. It bounced, then ground lower for most of the day. The RTH low broke at 10:30, and at 12:55 price was around 27,276, about -2.7% on the day. ES and RTY were also down, and VIX was up ~24%.
The trade:
• Short 2 MNQ @ 27,520 at 8:16 (the limit was 27,515, filled 5 pts worse). The bot's reasons: a failed bounce under the London low, a lower high, and a 5m close back below VWAP, with Daily/4H/1H all pointing down. It graded the setup B+. Stop 27,603, targets 27,383 / 27,340.
• Exit: stopped at 27,603 around 8:22. A fast rip took back VWAP and flipped the 1m gaps against the short. The bot's re-grade rule said to take half off, but the stop hit first.
• Then a bug in how my bot handles orders: its "take half off" buy went in after the position had already closed, so it opened a 1-lot long @ 27,606.75 by mistake. That was flattened by ~8:25 for about -$33.50.
• Result: -$365.50 in total, -1.10R.
Besides that trade, the bot set five resting short limits during the session (27,520, 27,450, 27,485, 27,340, 27,336). None of them filled. The 7:31 one missed by 1.25 pts, and the rest were pulled under their cancel rules. It read the direction right but never got filled on the pullbacks.
Running total since Sep 28 (one combined record; some early trades were counted as shadow fills while paper order placement was being fixed):
• Total P&L: +$1,738.00
• 24 trades, 13 wins: 54.2% win rate
• Profit factor: 1.39
• Current drawdown: -$642.50 from peak
• Max drawdown: -$1,335.50
The lesson: getting the direction right isn't enough. With the 5m ATR around 75, a short at VWAP had about one ATR of room, and one rip was enough to stop it out. Also, any order meant to reduce a position has to check that the position still exists before it's sent. That fix is on the bot's list.
#NQ #futures #daytrading #AItrading
80
Last week Blackwell made $1,363.25. Today we gave back $832.90 of it. 📉
Not exactly how I wanted to start the week, but this is what building and running an automated trading desk actually looks like.
Blackwell Capital | November 16
🔴 Net P&L: −$832.90
📊 4 trades | 1 win, 3 losses
🎯 Win rate: 25%
Today’s breakdown:
🔴 EMA Trend (MNQ): −$386.10
🟢 EMA Trend (MNQ): +$606.40
🔴 Williams %R (MGC): −$509.60
🔴 Morning Flush (MES): −$543.60
Interestingly, EMA Trend actually finished positive at +$220.30 net.
The losses came from our gold and Morning Flush strategies.
But the P&L isn’t the only thing that caught my attention.
We found discrepancies between the alert prices and actual broker fills, particularly on the gold and MES exits.
That doesn’t automatically mean slippage. Some exit payloads contain questionable reference prices, including one reporting a quantity of zero even though the broker correctly closed all five contracts.
That’s something we need to investigate before drawing conclusions about execution quality.
On the positive side, all eight received alerts were processed, all four positions were closed, and no failed exits were recorded.
Execution confirmations also remained fairly consistent, with exits locally confirmed in roughly 1.1–1.2 seconds. Exact broker latency still needs proper clock synchronization.
Our regime detector is now at 48 usable signal annotations across 9 strategy labels, but remains shadow-only while we work on validating its usefulness.
We’re still in drawdown, and today certainly didn’t help.
Last week’s profits weren’t proof we’d turned the corner, and today’s losses don’t automatically mean the strategies are broken.
What matters now is investigating the discrepancies, protecting the remaining drawdown cushion, and making decisions based on evidence rather than reacting emotionally to one session.
A red day, but the system completed its trades and finished flat.
Building Blackwell in public means showing days like this too. 🏗️
$MNQ $MES $MGC
31
Down 2% the day before Nvidia reports isn't a signal. It's a range warning.
NQ's overnight range alone was about 725 points. That's $14,500 on one mini. $1,450 on one micro.
If your stop was built for a quiet week, it's sized for a market that isn't here.
The fix: set the stop for today's range first, then cut contracts until the dollar risk matches a normal day.
Tomorrow: normal size, half size, or flat until after the print?
22
Your daily loss limit isn't a dollar number. It's a points number.
$1,000 daily loss on NQ:
1 mini = 50 points
2 minis = 25 points
5 micros = 100 points
10 micros = 50 points
Now put your normal stop next to it. A 20-point stop on 2 minis is $800. One loss and you've got $200 left for the day.
Quick check: daily limit ÷ your stop in dollars. If it's under 3, you're sized for one bad trade, not a day.
Save this. What's your number?
26
njoii.eth retweeted
OK, now that the market is closed, can we have a discussion?
🛑Stop full porting the U.S. market open.
It's a bad strategy. And odds are that it's not going to work out for you.
This week, we ended our funding relationship with 61 mostly new traders who got into PRO accounts and then decided to full port (e.g., trade large size) at the 9:30 AM ET open. This is a very small fraction of traders, but they're always consistently there. And I don't understand why.
Best case: they make a few thousand dollars. Worse case: they blow their PRO accounts. But either way, it ruins their chances with trading one of the best live-market prop firm programs out there.
The goal of prop firms is to turn good trading into good traders. In other words, to give you the experience and enforce discipline that will help you turn those stretches of good trading into a consistently profitable approach.
From a risk management perspective, trading full size at highly volatile times is not a strategy that will work out. If you do this, you have a 100% chance of ruin. So, it's not the key to a long-term relationship where we can win alongside of you.
First, I should be very clear: trading at the open isn't against our rules. Most traders do this responsibly inside their account and never hear from the risk team. A few do this irresponsibly. For them, we don't claw back profits (if they are lucky enough to make them). We simply identify the behavior, and notify traders that they're not a fit for our program moving forward. They're allowed to withdraw any profits they made, finish out their path in a PRO+ Development account, and then, we part ways.
But these traders aren't entirely irrational, are they? They have a 50/50 chance of winning a few thousand dollars, right?
The data suggests otherwise. Here are the stats from this week:
Monday: 3 of 11 traders made money
Tuesday: 3 of 13 traders made money
Wednesday: 4 of 8 traders made money
Thursday: 2 of 14 traders made money
Friday: 8 of 15 traders made money
In total, 20 of 61 traders made money. 33%.
Why is it so low? The issue with placing huge sized bets at a volatile time is that NQ might go up 15 points and then down 30 points in a few seconds. You can be wrong betting both ways if you try to get too much out of the trade.
But I'm sure you're much more disciplined than the traders above, and you put a bracket trade down so that you either hit it or don't: 15 NQ points – to match the 15 points of drawdown you have at max size. That's truly a 50/50 chance. OK, I'll agree. Now, let's say now that you guess right and hit your profit target.
Great, you just made $4,500 across five accounts. $22,500, right? What's the math behind that now?
You're still inside the buffer, so after the profit split, you just made $2,250 x 5 = $11,250 you can withdraw today. Not bad for 20 seconds of work. Don't forget to save for taxes (e.g., -25%). And you probably paid some to get into the accounts in the first place. So let's call it profits of $7,500… or $1,500 per account. Maybe that's worth it. But when you consider you just burned an opportunity to be a funding partner with a top firm in the space, what's it really costing you?
Or, instead of withdrawing, you now get to move into live with 5x PRO+ Development accounts. You either make these accounts work or go looking for a new funding partner. That's some pressure.
Look, I get it: the allure of full porting is real. Futures trading presents you with an opportunity to make thousands in seconds. But before you do that, do the math as to the real costs and benefits. Even if it works out, I'm not sure it pays.
The most dangerous trade next week is the first one on a fresh account.
A lot of evals and funded accounts got blown this week. Monday, the reset feels like a clean slate.
It isn't. The drawdown is at its tightest, and you're trying to win back the fee.
Day one on a new eval or funded: normal size, half size until there's a buffer, or skip the first day?
24
My AI discretionary trading bot: Fri Nov 13 recap (NQ / MNQ, TradingView paper, times PT)
Quick context: the old challenge experiment is stopped, but the learning keeps going with real paper trades every day. No P&L goal either way, just tracking how the bot does over time, P&L-wise and win/loss-wise.
Today: 1 trade, 1 win, +$108.00 (+0.37R).
What the market did:
• The open flushed hard. From the 28,399.25 high, NQ sold through the overnight low and printed a 27,690.5 low in the 8:20 bar.
• My bot sat that drop out. A buggy 1.5x ATR stop check combined with spiking volatility pushed risk over the bot's $400 cap even at 2 micros, so those shorts stayed watch-only. One short it marked near 6:45 ran ~160 pts, but the pre-news blackout ahead of UMich kept it out.
• After the low, NQ chopped in a range with 27,951 as the lid.
The trade:
• SHORT 3 MNQ. At 9:44 the bot rested a sell limit at 27,908 after the third rejection under 27,951. It filled at 27,920.75 during the 9:45-9:47 sweep to 27,954, 12.75 pts better than the limit. Stop 27,970, 1R = 49.25 pts.
• Why: range day, third failed push at the lid, price under VWAP.
• By 10:09 it was up about +0.75R. At 10:11 the bot moved the stop to 27,896 to lock +0.5R. It stopped out in the 10:15 bar at ~27,902.75 (6.75 pts of slippage). Result: +$108.00.
At 10:35 NQ closed above 27,951 for the first time and broke out to 28,049.5 by 11:05. The midday rule (A- or better from 10:00 to 11:30) kept the bot out of the breakout long. It was flat by the 12:50 cutoff, with NQ near 28,027.75 at 12:58.
Running total since the desk started Sep 28 (one combined record; some early trades were counted as shadow fills while paper order placement was being fixed):
• Total P&L: +$2,103.50
• 23 trades, 56.5% win rate
• Profit factor: 1.51
• Current drawdown: -$277 from peak
• Max drawdown: -$1,335.50
Lesson: the early profit lock paid for itself today. The range lid broke 20 minutes after the bot moved its stop. With the stop still at 27,970, this short would have been a full -1R loss. The +0.5R lock made it a small win instead. The trade-off was the morning: the big trend move came while volatility was too high for the bot's size rules, so it watched. That was the bot's call. Sitting out beats taking an oversized trade.
#NQ #futures #daytrading #AItrading
1
121















