Each candle is the story of one time slice: the body spans open→close, the wicks mark the high and low. Green = closed above open, red = closed below. Long wick + small body = a fight where one side got rejected. One candle means little — where it prints (at VWAP, at a band, after a trend) is everything.

A moving average that weights recent prices more. EMA9 = fast (reacts quick), EMA21 = medium, EMA50 = the bigger trend. When the fast crosses the slow, momentum just flipped. How your bots use it: the scalper's trigger is literally the EMA9/21 cross — but only when 6 other filters agree, because a naked crossover strategy backtested at a LOSS (-$362). The filters are the edge, not the cross.

Two exponential moving averages define the trend on each timeframe: EMA20 above EMA50 = UP, below = DOWN. Simple, but the bot checks it on two timeframes (H1 and H4) and only trades when both agree — a conflict costs 3 of its 10 confidence points, which almost always means HOLD. The best entries are pullbacks: price dipping back to EMA20/50 in a trend, not breakouts far from the averages (chasing costs a point).

A pullback is price returning to its moving averages inside a trend — you buy the dip in an uptrend, sell the bounce in a downtrend, with a tight stop just beyond the average. A chase is buying after price already ran far above the averages: your stop has to be wide and the snap-back comes for you. The bot measures distance to EMA20/50 against ATR; within ~0.6x ATR counts as a pullback.

The stock bot reads trend off daily simple moving averages: price above SMA20 above SMA50 = uptrend (the only thing it buys), the reverse = downtrend, anything tangled = "mixed" (hold). Stacked averages mean short-, medium-, and long-term buyers are all in profit — there's no overhead supply rushing to sell into rallies. When the stack breaks (price closes below SMA20 that is rolling over), the thesis is broken and the bot considers a SELL.

The average price paid today, weighted by volume — the institutions' benchmark for "fair value." Above VWAP, buyers who bought the average are in profit (bullish); below, sellers control. How your bots use it: the scalper only longs above / shorts below VWAP, must be at least 0.12x ATR away (out of the chop zone) but no more than 1.0x ATR (chasing an extended move mean-reverts into your stop — the exact bug that cost v2 money).

Volume-weighted average price for the session: the average price every share actually traded at today. Above it, the average buyer is winning and dips get bought; below it, the average buyer is trapped and rallies get sold. Institutions benchmark their fills against it. It only exists intraday — on the daily chart it's hidden.

Momentum on a 0–100 scale: how strong were recent up-moves vs down-moves. Above 70 = overbought (stretched up), below 30 = oversold. It does NOT mean "instantly reverse" — strong trends stay pinned overbought for a while. How your bots use it: as a quality filter, never a trigger. Futures: the hourly brain refused a short at RSI 26.8 (selling the bottom). Forex: buying above 70 or selling below 30 costs 2 confidence points; 40–60 is the sweet spot. Stocks: the bot wants bullish momentum with room left, not a buy at 75.

The gap between EMA12 and EMA26 (the MACD line), plus an EMA9 of that gap (the signal line). Line above signal = momentum building up; the histogram shows the gap growing or shrinking. It lags a little but filters noise. How your bots use it: confirmation. The hourly futures brain and the forex brain both require MACD to AGREE with the trend direction before taking a setup; disagreement is a deduction, not a veto.

The average size of a bar, including gaps — the market's volatility ruler. ATR doesn't say direction; it says how far things move, which sets stop distance and position size. Futures: scalper stop 1.4x ATR / target 3.0x; hourly bot 1.5x ATR stops, 2.5:1. Forex: the bot fixes 15–25 pip stops — when ATR is bigger than the stop, one ordinary bar can stop you out. Stocks: a fixed -3% stop on a name whose daily ATR is 4% is a coin flip. Size shrinks as volatility grows; that is the whole point.

Trend strength (not direction) on 0–100. Under ~18 the market is chopping — crossover signals fire constantly and lose constantly. Over 25 = real trend. How your bots use it: the scalper's v4 regime gate refuses any entry with ADX below 18. Lesson learned the hard way: 7 of its first 8 losses came from trading the 9:30 open chop.

A 20-bar average with bands 2 standard deviations out — ~95% of closes fall inside. Touching a band = stretched; tight bands (a "squeeze") = energy loading for a breakout. How to use them: in chop, fade the bands back to the middle; in a trend, price can WALK the band — so check trend strength (ADX, stacked averages) first.

A pip is the 4th decimal on most pairs (0.0001) and the 2nd on yen pairs (0.01). Trading 1,000 units, a pip is worth about $0.10 — the paper bot's 5,000-unit positions make $0.50 per pip, so a 20-pip stop risks $10 and a 40-pip target makes $20. Sessions: London (3am–12pm ET) and New York (8am–5pm ET) bring the volume; the overlap (8am–12pm) is the best two hours of the day. Asia is thin; the bot keeps size small there. Forex is closed from Friday 5pm to Sunday 5pm ET.

It starts every setup at 10/10 confidence and deducts: H1 vs H4 trend conflict −3, RSI stretched against the trade −2, MACD disagreeing −1, chasing far from the EMAs −1. Below 7 the bot holds. Then the hard rules: stop 15–25 pips, target at least 2:1, risk 1% of the balance per trade, one position per pair, and a trailing stop that ratchets behind price so a winner can't turn into a loser. Most hours it does nothing — that is the strategy working.

Every buy goes in as a bracket: the entry plus two exit orders attached at once — a stop 3% below and a limit target 6% above. Whichever fills first cancels the other. That is a 2:1 reward-to-risk before the trade even starts, decided while the bot is calm, not while the stock is moving. The bot never adds to a position, caps any single stock at 10% of equity, and only sells early when the trend thesis breaks — never for sizing housekeeping.

The math that keeps you alive: risk a fixed small amount per trade and make winners bigger than losers. At 2.5:1 reward-to-risk you're profitable winning just 30% of trades; at 1:1 you need 55%+. Your real settings: $667 max daily loss (1/3 of the TopStep $2,000 floor), $133 max per trade (so 5 straight stops can't end the day), 2.5:1 targets, flat before news events and before the close. Amateurs blow up on size, not on entries.