I am not well versed on the current iteration of AutoSpreader - but several years ago, when one spread leg was filled you had the option of using a limit order to fill the remaining leg. I really don't need AutoSpreader because I am swing trading.
Some time back I designed a chart tool for Ninjatrader8 that trades futures spreads, can scale in-out and has a money management system for SL and TP with fixed dollar amount for SL. The slippage and commissions add up with too many trades, so its necessary to limit no of trades per day. The slippage issue remains whether its going to be market orders on both legs or limit on one leg and market on other. Initial 30 min range breakout was the strategy I am testing with this tool. Autospreader like TT/CQG is absolute must if anybody is attempting intraday spreads on equity futures. Autospreader cost is expected to compensate for the slippage in filling, somebody who has used this in live trading may confirm. About the intraday movement of spreads, when NQ outperforms, ES too does relative to YM as tech contribution is more in ES than in YM. RTY or small caps has altogether random moves, poor correlation with YM or NQ, so ES/RTY may be only relevant spread with RTY. Of course the big 5-6 FAAMGs need to be watched when NQ is one of the components of spread trade. YM has 20% tech weightage while ES has 25%.
Hello - Does anyone here have experience with TT Pro Autospreader for ES/NQ, ES/YM, ES/RTY spreads? Thx
Its a simple detrending script. You will have to make the studies in thinkScript. Set it up like this, with the left axis checked for the overlay and the Relative Position study on the lower list. Overlay() Code: input Ticker = ""; plot X = close(Ticker); LinearRegCurve(0,150,CLOSE) Code: input displace = 0; input length = 150; input price = close; plot LinReg = Inertia(price[-displace], length); LinReg.setDefaultColor(GetColor(1)); Relative Position (0,150,CLOSE) Code: input displace = 0; input length = 150; input price = close; plot LinReg = Inertia(price[-displace], length); LinReg.setDefaultColor(GetColor(1)); plot X = price - LinReg; plot Fast = Inertia(X, 20); plot Slow = Inertia(X, 50); plot Z = 0;
Thanks for all the great information about spreading the MES / MNQ! When I look at the actual book from MES and MNQ the prices move very quickly in and around the b/a, there's no way to catch them without full automation. IBKR's python and C/C++ API might be fast enough. Haven't tested it, live. IBKR's quotes for a 3MES/2MNQ are bizarre and move wildly, there is No way to get this executed at reasonable levels, let alone any liquidity concerns. 2 questions - anyone try automating this with the IBKR API? - anyone investigated CME colocation to execute? I was investigating how to trade them as a cointegrated pair to trade the spread based on the z-score of the spread. It looks possible but the fees, commissions, spreads it looks iffy. Maybe automating a regular trading system is more cost efficient?
First of all, a Micro's spread like this isn't leverageable from a fees perspective. You pay more in exchange/commision on micros as a percentage of notional exposure. The groups making and taking liquidity (short time frames) are definitely co-located/exchange members/market makers etc. Also, you really can't compare the b/a spreads during RTH to what's quoted in the outside hours. After hours and premarket spreads are wide as fuck, and that goes for any listed index spread on CME. The spread you quoted is a hedge (synthetic index/synthetic future) for a large cap portfolio that's short the high beta and long the lower beta index constituents (though since both contracts are cap weighted, the index weightings are working against this). In other words, your long leg has structural beta exposure (due to cap weighting) but you hedged out tech exposure with the short (in-fact over hedged). This spread doesn't really make sense and its actually bearish short term, but long term had a hard rally. You essentially have a modified beta risk which is trading not on cointegration statistics but on index vol and general market moves.
Thanks RealMoney, you're right to hedge properly really need to approach it from risk perspective. Cointegration just looks at the stationarity of the time series. This thread has saved me time writing a bunch of code to automate it. I'll put my time and effort into automating a directional strategy, maybe take a look LSTM's for a classification algo, or some of the new Time Series ML models. Test, test, test. Thanks again!
Hello GreatRedOne99 As you observed, MES/MNQ spreads are wider than ES/NQ. ES is decent but NQ can be a bit wild at times. TT Pro has their servers hosted near CME and TT Auto Spreader has advanced logic in spread order handling. I think CQG Spreader is similar. Market order with MES/MNQ is a challenge with intraday trading of this spread. Market order with ES/NQ can work most of the time (because you are going long on one while going short on the other, while 2 move in sync) but high volatile/news events can skew the bid/ask spread. Limit order using TT or CQG Spreader executed by the colocated servers is an edge in this case. Thanks much to RealMoney for explaining the validity and short term/ long term bias with trading ES/NQ pair.