Grid vs DCA: The Math Behind Two Bot Strategies
Both grid trading and Dollar Cost Averaging promise consistent returns. The math says one is the right tool only in specific market regimes. Misuse the other and you compound the losses.
What grid actually does
A grid bot places buy orders at intervals below current price and sell orders above. Every fill triggers a counter-order. In a lateral market, this captures volatility — small profit on each round trip.
The math is precise: profit per cycle = price step × quantity − fees. With BTC at $80,000, a 0.5% grid step, and a $200 order size, each completed round trip nets ~$0.94 (after 0.10% taker × 2 fees). 100 round trips/month = $94 on $200 deployed = 47% annualized.
Caveat: ONLY in lateral. If price breaks out of grid range, you accumulate the wrong side and watch the other go.
What DCA actually does
DCA buys at regular intervals (time or price-based). Original retail tactic: buy every Friday regardless of price. Modern DCA bots add safety orders at percentage drops — averaging down on dips.
The math: average entry approaches a weighted mean. In a sustained downtrend, your average keeps falling but you're never recovering — you're buying losses. In a sustained uptrend, DCA underperforms lump sum because you missed the early entries.
DCA wins in: choppy markets with strong long-term trend, where dips are bought and the trend rescues the bag. It loses in: prolonged bear markets, parabolic blow-offs.
The regime test
Before deploying either, ask: what's the volatility regime, and what's the trend?
- Lateral + low vol: grid wins. DCA gets bored.
- Lateral + high vol: grid wins big. Each swing pays.
- Trending up: DCA OK if accumulating. Grid loses (sells too early).
- Trending down: NEITHER. Both compound losses.
What we actually run
TradingIA runs grids on BTC, ETH, SOL, AAVE — assets with proven mean-reversion in 7-15% bands. We DCA only assets we're long-term bullish on, with hard stop on draw-down.
The combination: grid harvests volatility while we hold the spot bag. Funding income (when positive funding) adds another layer. The result is risk-parity at the strategy level — no single regime breaks the system.
The retail mistake
Retail picks one bot, deploys on whatever asset is trending, and assumes "it just works". The bot reflects the market. If the market changes regime, the bot loses — predictably.
Pro shop: pick the bot for the regime, not the regime for the bot.
A grid in a downtrend isn't broken. The grid is doing exactly what it's supposed to. The user picked the wrong tool for the regime. Tools don't fail. Choices do.