Przejdź do głównej zawartości

Transaction Cost Analysis (TCA)

Execution alpha to tiny per trade i enormous over thousands z nich — to large part z jak banks i prop desks keep ich edge. TCA measures jak daleko price którą rzeczywiście achieved drifted z price gdy zdecydowałeś do trade.

Otwórz cBots → Execution Cost (/quant/tca).

Co measures​

Biorąc arrival (decision) price, side, i Twoje fills (price × quantity), to reports:

  • Average fill price (VWAP) — volume-weighted price którą rzeczywiście got.
  • Slippage (bps) — drift z arrival do VWAP w basis points, signed więc positive number to cost (buying above arrival lub selling below to) i negative number to price improvement.
  • Implementation shortfall — ten cost expressed w price × quantity terms: pieniądze które drift cost na ten order.
POST /api/quant/tca
{ "arrivalPrice": 1.1000, "side": "Buy",
  "fills": [ { "price": 1.1010, "quantity": 100 }, { "price": 1.1020, "quantity": 100 } ] }

Smart slicing (Almgren-Chriss)​

Beyond measuring cost, cMind może plan large order do minimize to. cBots → Execution Schedule (/quant/execution) builds Almgren-Chriss optimal-execution schedule: biorąc total quantity, number z slices, Twoja risk aversion, volatility i temporary market impact, zwraca size do trade w każdy slice. Wyższy risk aversion front-loads schedule (cutting timing risk); zero risk aversion flattens do even TWAP. Slices zawsze sum do total.

POST /api/quant/execution-schedule
{ "totalQuantity": 100, "slices": 5, "riskAversion": 2, "volatility": 0.02, "temporaryImpact": 0.1 }

Dlaczego jest niezawodny​

Pure, deterministyczne domain code (Core.Execution) z żadną infrastrukturą dependency i no external calls — unit-tested dla buy/sell cost sign, price improvement, zero-slippage, VWAP aggregation, i input guards. To measurement half z execution quality; to jest same shortfall metric które copy engine uses do judge (i, z smart slicing, reduce) cost z mirrored orders.