The market gave us 21 breakouts today. The AI bots caught zero of them.
A brutally honest recap of what happened, what we missed, and what we fixed before tomorrow.
The market gave us 19 breakouts today. The bots caught zero of them.
A brutally honest recap of what happened, what we missed, and what we fixed before tomorrow.
QQQ led the session with tech carrying the tape. SPY held green but breadth was narrow. The scanner flagged 51 setups across 29 bulls and 22 bears, but quality was thin: only VRT and UBER graded A. The remaining 42 A- setups lacked the confluence needed for high-conviction entries.
By the close, 19 setups resolved as breakouts, 28 failed, and the rest expired. The story of the day was the intraday reversal: SPY opened strong but rolled over below VWAP around 10:45am, dragging mega-caps lower and creating a cascade of BEAR breakouts the bots weren't equipped to trade.
UBER was the cleanest A-rated bull with 74 pressure, 1.95x RVOL, and full intraday trend alignment across the 5m, 30m, and 4h timeframes. It sat only 0.54% below the $75.56 breakout level, making it the closest high-quality trigger among the top setups.
What happened after the pick
UBER never triggered. Despite being 0.54% from its breakout level at the morning brief, the intraday reversal pulled it away from the level as SPY weakened. By 11:00am, UBER was trading at $74.01, down 1.22% on the day. The BULL thesis was intact on higher timeframes, but the intraday action killed the trigger window.
Arxe's hindsight: The pick was directionally correct on the daily chart, but the intraday reversal made the morning window untradeable. A better framework would have flagged UBER as a swing hold rather than a day trade trigger. The backtest shows UBER has an 85.7% breakout rate across 7 scanner appearances with a 5.03x profit factor. The setup will likely return.
UBER in the backtest
UBER is one of the strongest performers in the 188-symbol universe. Across 423 backtested trades, it ranks in the top 5% by profit factor. When this setup reappears, it deserves the first slot on the watchlist.
How the setup evolved today
UBER appeared in the scanner at 9:31am with an A- grade, upgraded to A by 10:11am as pressure built and RVOL climbed from 0.39x to 1.95x. The setup sat at "On Watch" (0.54% from breakout) for most of the morning before the intraday reversal pushed it away.
| Time | Grade | Pressure | RVOL | Dist | Conf | Trend |
|---|---|---|---|---|---|---|
| 3:51pm | A- | 74 | 0.85x | 0.55% | 9/10 | ▲ |
| 3:41pm | A- | 77 | 0.42x | 0.62% | 9/10 | ▲ |
| 3:31pm | A- | 77 | 1.28x | 0.67% | 10/10 | — |
| 3:22pm | A- | 79 | 0.96x | 0.65% | 10/10 | ▲ |
| 3:11pm | A- | 79 | 0.39x | 1.17% | 9/10 | ● |
What the bots did today
Wex — The Edge Hunter
Wex built a 5-setup watchlist throughout the day: TXN (#1), SPOT (#3 BEAR), DIS (#4), UNH (#5), and AMZN (#5). Earlier in the morning, IWM, GOOGL, and XLE were also on the watchlist but were dropped during intraday refreshes as the market reversed.
None of the 5 final setups triggered. TXN had the highest pressure (87) but sat 0.5% from its breakout level. SPOT was a BEAR setup that made it onto the watchlist via the new intraday SPY reversal logic, but it never reached its breakdown level either.
The IWM miss: IWM was on the watchlist at 9:35am, ranked #3 as a day trade with medium conviction. It sat at breakout ($280.58) with 1.2x RVOL and 76 pressure. But the plan_entry was set at $281.02 (0.5 ATR buffer), and IWM's price never pushed that far. The human trader entered at $0.33 and rode it to $0.49 (+48%). After this miss, we lowered the ETF entry buffer from 0.5 ATR to 0.25 ATR for mega-liquid ETFs. Tomorrow, Wex would have triggered on IWM.
Wex account: ~$6,853 (started $10,000). 7 closed trades, 0 wins. All losses to date have been infrastructure bugs, not strategy failures.
Xcel — The Accelerator
Opened at 9:35am, the first trade of the day. USO had 8.45x RVOL (oil sector was active) and A- grade with 61 pressure. GPT approved the entry. Xcel bought 12x $142 calls at $4.85 (2-DTE).
The problem: USO barely moved. Price went from $142.90 to $143.25 (+0.2%) but the $4.85 premium dropped to $4.57. GPT cut it after 22 minutes for "barely moved despite favorable conditions." Small loss, disciplined exit.
Arxe's take: USO's RVOL was inflated by the opening surge. At 9:35am, cumulative RVOL always looks extreme because only 5 minutes of volume have accumulated against a full-day average. The time-aware RVOL fix (deployed today) would have flagged this as less impressive than it appeared. The $4.85 premium was also too expensive. With the new cheaper-OTM contract selection (also deployed today), Xcel would have picked a $0.30-$0.50 contract with more leverage.
Opened at 10:20am. ETN had 1.75x RVOL and A- grade. But there was a critical bug: the DB recorded entry at $13.46 (the mid-price estimate) while the actual Alpaca fill was $17.40. GPT thought the trade was up +21% when it was actually down -6%. It scaled out at what it believed was a profit, locking in a real loss.
GPT finally cut the remaining position after 43 minutes at $14.70, down -15.52% from the real entry.
Arxe's take: This was the entry price bug in action. Every GPT evaluation was based on phantom P&L. The $17.40 fill on a $13.46 mid-price means the bid/ask spread was enormous, 29% above mid. This is exactly the kind of illiquid contract the new delta range (0.25-0.45) and cheapest-first sort would have avoided. Additionally, ETN had only 1 backtest trade (a loss). Not a high-conviction name.
Opened at 10:40am. ON had 1.73x RVOL, A- grade, 73 pressure. Xcel bought 6x $105 calls at $6.55 (11-DTE). GPT cut it after 21 minutes at $6.03 as the stock stalled at $103.40.
Arxe's take: ON actually has strong backtest numbers: 66.7% win rate, 1.99x profit factor, +0.33R avg across 6 trades. The setup was legitimate. But the 11-DTE contract at $6.55 was far too expensive for a day trade. With 0-1 DTE at a delta of 0.25-0.40, Xcel would have paid ~$0.80 and the -7.94% loss would have been a -$48 loss instead of a -$312 loss. Same outcome, much smaller damage.
Opened at 10:50am, the last trade of the day. IWM had 1.19x RVOL, A- grade. Xcel bought 14x $281 calls at $1.04 (1-DTE). GPT cut it after just 10 minutes at $0.93 as IWM pulled back to $279.82.
Arxe's take: Xcel entered IWM an hour after the human trader, at a higher premium ($1.04 vs $0.33), and at a different strike. By 10:50am the breakout was already fading. The human caught the initial push at 9:35am; Xcel caught the pullback. Timing is everything with 0-1 DTE options. The 10-minute hold before GPT cut was actually too fast. Breakouts need 15-30 minutes to develop (the SKIP_FIRST_BAR lesson applies to exits too).
What unXcellent caught that the AI bots missed
Entered at 9:35am. $281 calls at $0.33 (0DTE). Rode to $0.49. Saw the setup on /proximity at breakout with 11/11 bias, 10/10 confirmations, 1.73x RVOL. Trusted the level and took the trade.
Entered around 10:30am. $195 puts at $0.39 (weekly). Rode to $1.11. Saw SPY/QQQ breaking down intraday and NVDA leading the downside with 2.0x RVOL. Bought puts on the support breakdown.
The scanner had NVDA graded B+ because the 30m and 4h trends were NEUTRAL (hadn't caught up to the move). The bots couldn't see it because spy_trend was still BULL. The human read the intraday context that the algorithm hadn't encoded yet.
Lesson: The scanner is a pattern detector. The human is a context reader. Today's NVDA trade triggered three fixes: intraday SPY trend (so bots can short when SPY sells off intraday), momentum grade override (so extreme RVOL + RSI boosts the grade), and ETF entry buffer reduction (so IWM triggers faster). All three are live for tomorrow.
19 breakouts, 28 failures, 16 expired
The scanner resolved 63 setups today. The breakout rate was 30.2% (19/63), which is below the backtest average of 58.1%. This is expected on a day with an intraday reversal: many BULL setups that were graded A- in the morning failed as the market turned.
Notable breakouts
Six BEAR A- setups broke out for profit today. All six had spy_trend = BULL, which meant the bots couldn't trade any of them. This was the #1 gap identified today, and it's now fixed with the intraday SPY trend override deployed to both bots this afternoon.
Energy is the only sector outperforming SPY over 5 days. Gold is getting crushed (-8.6%), which aligns with GLD and GDX breaking down today. Biotech and housing continue to lag. If you're looking for BEAR setups this week, housing and gold miners are where the weakness is concentrated.
13 changes deployed across 4 services
Today was the most significant system upgrade since launch. Every fix was driven by a specific trade that was missed or botched:
Scanner (5 changes)
Intraday SPY trend (VWAP-based, not just 200 EMA). Time-aware RVOL (compares to same time of day). ETF RVOL tiers (SPY/QQQ/IWM almost never filtered). Momentum grade override (B+ to A- on extreme conditions). Grade stability check (prevents flickering).
Xcel (5 changes)
Entry price bug fixed (polls Alpaca position for real fill). MAX_CONTRACTS lowered to 10. Delta range narrowed for cheaper OTM contracts. GPT entry hierarchy (ETFs always approved). BEAR blown-past widened to 1.0%.
Both Bots (3 changes)
Intraday SPY trend BEAR override. ETF entry buffer lowered to 0.25 ATR. Arxe gate integration (advisory mode).
The system works. The execution is catching up.
Across 423 backtested trades on 156+ symbols, the wedge breakout algorithm shows a consistent positive edge. The backtest improved from 55.9% to 58.1% WR after the SKIP_FIRST_BAR discovery (opening-bar entries caused all negative expectancy). Today's live performance doesn't match the backtest yet because the bots are still working through infrastructure bugs and filter calibration. Every bug fixed brings the live results closer to the theoretical edge.
What to watch tomorrow
Tomorrow (Tuesday, May 5) will be the first trading day with ALL upgrades active: intraday SPY trend, momentum grade override, cheaper OTM contracts, accurate entry prices, and the Arxe gate. If the market reverses intraday again, the bots will now be able to take BEAR trades for the first time.
Key setups to monitor: UBER remains the highest-conviction name with 85.7% backtest win rate. TXN (71.4% WR, 2.49x PF) is coiling at resistance. If energy continues leading, XOP and OXY are at breakout levels. On the BEAR side, GLD and gold miners (GDX, GDXJ) are breaking down with sector momentum behind them.
IBM reports after the close today. Watch for any scanner-universe names that gap on sympathy moves tomorrow morning.