Chapter III · How the system works

From thousands of stocks to one disciplined decision.

TradeWise is an AI trading platform that runs a team of nine strategy bots. Every bot shares the same foundation: a scanner that chooses what to watch, live price data, a backtester that tests every rule, a broker that executes, and a record of every step.

  1. Nightly · 4:20 PM CTOpportunity ScannerChooses what each bot watches
  2. Market hoursStrategy botsEach trades its own tested strategy
  3. BrokerTradeStationSupplies prices, executes orders
  4. Every tradeAlerts and journalDiscord message via Peakbot plus a full record

Step 1 · The Opportunity Scanner

Every night, the owl picks what to watch.

After each weekday close, the scanner reviews 5,000+ US stocks: the S&P 500, the Russell 1000 (adding large and mid-caps that aren't in the S&P), the Russell 2000, and the total US market (ITOT), plus our own custom list. The bots switch to the new list at the next open. The scanner never places orders; a stock on the list still needs a confirmed signal.

Stage 1 · Quality filter

  • Price of at least $5
  • Custom volume shares traded per day on average per trading bot
  • A weekly up-trend: price above a rising 30-week average

Stage 2 · Ranked for the breakout bots

  1. Pause in progress: broke out, held, and is close to its trigger price
  2. Near a breakout: within 3% of its 20-bar high
  3. Other up-trends: used only to fill the list
5,000+
stocks reviewed nightly
Top 300
watchlist per bot, plus current holdings
Every day
a scored report is kept to measure the ranking

Step 2 · The strategies

One platform. A team of strategy bots.

No single strategy works in every market. So instead of one bot, we run a team of nine, across three strategies, each built for a different kind of move. Every strategy is backtested on years of history and paper traded in SIM before it earns real money. Today the breakout bots trade live, the intraday bots and the ML AutoTrader trade in SIM, and new strategies are in development.

5 bots · Live

Breakout bots

Our original WiseTrader strategy on five charts, from 5-minute to 4-hour. They buy strong stocks in a weekly up-trend after a confirmed breakout, and can hold overnight.

3 bots · Paper trading

Intraday trend bots

Three daytime setups, a momentum cross, a red-to-green reclaim and a high-of-day break, each taken only when SPY is in an intraday up-trend and the stock is above its VWAP. Always flat before the close.

1 bot · In testing

Wise-Trader ML AutoTrader

Our first auto-trader, and the start of TradeWise. It scores S&P 500 stocks on 21 criteria, asks a machine-learning model how confident it is, and trades for a 1% gain per trade, hands-free from entry to exit.

In research

What's next

New strategies for different market conditions. Each one goes through the same path: backtest, paper trading, then a weekly check against its own backtest.

Strategy 1 · Breakout bots · The signal

Breakout. Pause. Trigger.

A stock must prove itself three times before it earns a Buy signal.

Recent high Trigger price 1 Breakout 2 Pause holds 3 Trigger: Buy Illustration only, not a real chart

1 · Breakout

The stock closes above its recent high while in a weekly up-trend.

2 · Pause

It holds above the breakout level instead of falling back.

3 · Trigger

It pushes past the pause high. That confirmed bar is the Buy signal.

Every signal also carries a risk count (warning signs such as a weak market) and a trend phase: Early, Established or Mature. An early, strong breakout is labeled Thrust.

Step 3 · The rules

Nine bots, each with its own tested rules.

The five breakout bots run the same WiseTrader signal, so a Buy means the same thing on every chart; what differs is how each one trades it. The three intraday trend bots each trade a different daytime setup on the same engine, with targets and stops sized to each stock's own volatility (ATR), and they are always flat before the close. The Wise-Trader ML AutoTrader scores and trades on its own. Every bot has its own take profit, stop, entry filter, volume floor and market rule, each chosen by testing.

BotStrategyTake profitStopEntry filterVolume floorMarket rule
TWS 5mBreakout2.5%10%Thrust + risk 0–4250K shares traded todaySPY below 20-day avg: Thrust only
TWS 15mBreakout1%7.5%Thrust + risk 0–21.5M shares traded todayNone
TWS 30mBreakout1%7.5%Thrust + risk 0–43M shares traded todayNone
TWS 1hBreakout1%7.5%Thrust + risk 0–11M shares traded todaySPY below 20-day avg: no new buys
TWS 4hBreakout1.5%10%Thrust + risk 0–3No minimumSPY below 20-day avg: no new buys
TWS EMAX 15mIntraday trend0.75 × ATR (about 0.9%), at a bar close10-bar low, capped at 3 × ATR9-EMA crosses above 20-EMA, close above VWAP, 11:00–14:00 ET1M shares and $50M a day avg; bar volume at least normalSPY above its VWAP with 9-EMA above 20-EMA
TWS R2G 30mIntraday trend0.75 × ATR (about 1.9%), at a bar closeLow of the day, capped at 3 × ATRWeak open, then first close back above yesterday's close and VWAP, 10:00–12:30 ET1M shares and $100M a day avg; bar volume 1.3× normalSPY green on the day
TWS HOD 30mIntraday trend0.5 × ATR (about 1.3%); stop moves to break-even after half the risk is earned10-bar low, capped at 3 × ATRHolds VWAP and closes above the high of the day, 11:00–14:30 ET1M shares and $100M a day avg; bar volume 1.3× normalSPY green on the day
Wise-Trader MLML AutoTrader1%[Stop %]21-criteria score + ML confidence[Volume]S&P 500 universe

Chosen by the backtester, not by guesswork

Our backtester replays the exact same indicator over past market history and simulates every trade the way the bots make it. We tested about 18 million combinations of take profit, stop, signal type, volume and market filter. For each bot we kept the settings that scored best and also held up when we nudged them slightly, and that were profitable in both the older and the newer half of the history. Every change is tested there first, before a bot ever trades it.

Backtests are simulated results chosen with hindsight. Live results will differ and are usually lower.

Before every buy

  • The Buy signal is one the bot accepts: Thrust, or a risk count within its limit.
  • Enough shares have traded today to meet the bot's volume floor.
  • The market filter allows it (SPY above its 20-day average, where the bot uses one).
  • The bar closed during regular market hours.
  • There's room under the bot's position limits.
  • No other bot owns the stock: each stock has one owner bot.
  • Size is a fixed dollar amount per trade, rounded down to whole shares.

How a trade happens

  • Entry: when a bar closes with a qualifying Buy, a market order goes out right away and fills at about the next bar's open.
  • Protection: the moment it fills, a stop order is placed at TradeStation. It stays there overnight, even if the bot is switched off.
  • Exit, whichever comes first: the take-profit target is reached, the indicator gives an exit signal at a bar close, or the stop at TradeStation is hit.
  • Manual close from the bot's control panel is always available.

Strategy 2 · Intraday trend bots

Trade with the market's trend, inside the day.

Three newer bots built on the same order engine, each trading one daytime setup drawn from professional trading playbooks. We turned the strongest ideas into rules, tested them on two years of history across hundreds of stocks, and kept the three that held up best. None of them holds a position overnight.

The three setups

  • EMAX 15m, momentum cross: the 9-EMA crosses above the 20-EMA with price above VWAP, in the late-morning to early-afternoon window.
  • R2G 30m, red to green: a stock that opened below yesterday's close reclaims it above VWAP on strong volume, early in the day.
  • HOD 30m, high-of-day break: a stock holding VWAP closes above its high of the day with a strong bar.
  • For all three: SPY must be in an intraday up-trend, the stock must be above its VWAP, and volume must be above normal for that time of day.

How they trade

  • Targets sized to the stock: the profit target is set in ATR, so a quiet stock and a volatile one each get a target that fits, and the stop is the recent low, capped at 3 × ATR.
  • Tested, then checked: settings were chosen on history before 2026, then checked on 2026 data the test had never seen, and only setups whose neighbouring settings also held up were kept.
  • Liquid names only: at least 1M shares and $50–100M traded a day on average, with a price between $5 and $500.
  • Same engine, same safety: the same broker connection, stop protection and safety rules as the breakout bots, with every position closed before 4 PM ET. All three are paper trading in TradeStation's simulated account.

Backtests are simulated results chosen with hindsight. Live results will differ and are usually lower.

Strategy 3 · Wise-Trader ML AutoTrader

Where TradeWise began: one percent, hands-free.

Before the breakout and intraday bots, there was the Wise-Trader. It is still running, and it is the most machine-learning-driven bot on the team.

How it picks a trade

  • Scan: it loads the S&P 500 and scores every stock on 21 criteria.
  • Ask the model: a machine-learning classifier, trained on past setups, gives each candidate a confidence score.
  • Trade the best: only the highest-confidence setups are taken, during the busiest market hours.

How it manages the trade

  • Target: a 1% gain per trade, the goal the whole company grew from.
  • Hands-free: entry, protection and exit are automated, with no emotion in the loop.
  • Broker: it trades through Alpaca, with a web dashboard that shows its signals and orders.

A 1% target is a rule the bot follows, not a promise of results. Trades can and do lose.

Step 4 · Safety

Built to fail safe.

Every strategy runs on the same order engine. When in doubt, the bots stop and check rather than send another order.

No duplicate orders

Every order is recorded before it's sent. If a reply is lost, the bot checks with the broker before trying again.

Always protected

Every filled buy, even a partial one, gets a stop order at the broker, so it stays protected if the computer goes down.

Clean exits

A sell converts the existing stop into a market order, so two orders never compete for the same shares.

Restart-safe

After a restart, each bot picks up its open positions and stops from the broker.

One bot per stock

A stock held by one bot is locked for the others.

Regular hours only

No after-hours orders, and market holidays are detected before the open.

Checked price data

The TWS Core Bar Hub streams prices from TradeStation once for every watched stock, and the breakout and intraday bots read from it. The raw data is quality-checked against TradeStation's own bars every day and reviewed again every week.

A full journal

Every buy, sell, fill and rejection is recorded with price, stop, result and order details.

Step 5 · Alerts

Nothing happens out of sight.

Every fill from the breakout bots sends a Discord WebHook via Peakbot. Each bot also has a desktop control panel showing its watchlist, open positions and settings, with one-click charts.

TWS FILLED BUY NVDA | 5m | 15 sh @ 210.90 | 09:35 ET
TWS FILLED BUY INTC | 15m | 30 sh @ 84.54 | 10:15 ET TWS FILLED SELL NVDA | 5m | Take profit | 15 sh @ 213.25 | +1.11% | 11:10 ET TWS FILLED SELL INTC | 15m | Take profit | 30 sh @ 86.15 | +1.92% | 12:30 ET

Sample messages with illustrative prices, not real trades.

Step 6 · Where we're headed

The future of trading bots.

Our vision is AI trading that grows by evidence: more strategies, tested harder, measured every week, and only trusted with real money once they've earned it.

A bot that grades itself

Strategy Health compares each bot's live results with its backtest every week and gives a verdict: OK, watch, half size or pause.

One core bot

We're building TWS Core, one bot that combines what we've learned on every timeframe to pick the best two to four trades a day, or none.

More strategies

New bots for different market conditions, each one backtested and paper traded before it joins the team.

In the cloud

We're moving the bots to a cloud server so they run around the clock and can be reached as web apps.

Where we are today

Tested before trusted.

Done

Dry run and SIM

Every bot first ran on real prices with no orders, then placed real orders in TradeStation's simulated account.

Today

SIM and live

The five breakout bots trade live with real money at small size. The three intraday bots and the ML AutoTrader are paper trading in SIM.

Next

More bots, more strategies

New strategies are in development. Each one earns live money the same way: backtest, SIM, then a weekly check against its own record.

When we share results, they will be real, dated and complete, losses included.

Next · Chapter IV

Follow the build

Join the waitlist →