How do you adjust for trading costs when backtesting technical analysis?

No matter the base of your strategy (TA, fundament, news events, options, AI), you must always account for all the costs, or your expectations are flawed. Here are some of them: Basic costs: broker commissions, exchange fees, slippages (executed price vs. expected trade price), taxes (duty fees, tax for financial transaction). Interest costs: interest (for margin trading, leverage, or derivative contracts), interest differential (when trading currencies), broker interest fee (yes, you pay it, and sometimes it is higher than interest - or it can destroy positive interest differential) Borrow-fees or restrictions for short selling, cost of data feeds, infrastructure cost, random error costs (wrong executions, failures, etc.)

When backtesting technical analysis, it's vital to factor in trading costs for a realistic assessment of your strategy's performance. Include brokerage fees, commissions, and slippage in calculations. Consider percentage-based costs, simulate slippage, and assess the impact of trading frequency. Account for tax implications, optimize considering transaction costs, and adjust capital allocation realistically. Evaluating net returns after factoring in all costs provides a more accurate representation of strategy viability in live trading.

We have found that in back-testing, to assume liquidity at the advertised spread is fundamentally flawed. For in reality the true market liquidity is as much 1% to 2% slippage for any entry or exit point of size greater then 1% of the daily market value of the traded security. Hence most advertised back-testing claims 🙄 have little credibility. This is especially true when trading equity or indicies options.

Leverage looks lucrative but it is a double edged sword. Always make calculations based on your risk tolerance before entering a trade. If you don't then it might be too late once you are already stuck in a trade.

There are two major ways: 1. Back test on real markets (Transaction charges applies). 2. Back test on Paper trading apps (Free). Real market trades includes charges while buying as well as selling. These charges are different for single lots and multiple lots Note them down. Make a journal at the end of day everyday. You are good to go! 2. Paper trading apps are completely free but the problem lies in lagging. Most paper trading apps don't give real time data. They lag by 1 minute or 2 depending on the app.

Estimating commissions and exchange fees is straightforward. Estimating slippage is also simple when you have bid-ask historical data. Still, most traders don't have it, so they can use volatility and dollar volume (low volumes have much higher slippages, same for high volatility). For intraday trading, forget stocks with less than 1 million USD average daily volume; only experienced professionals and algorithms can handle those slippages (+ limit orders only!) To handle interest, you can calculate it from base rates published by central banks (fed rates). Borrow fees are more tricky: I downloaded historical data from Interactive Brokers and used linear regression to estimate unknown ones (features: volume, volatility, fed rate)

Nowadays all brokers are up to date. When u hit buying option, they first show the charges that will be applied to the actual trading capital for that particular trade. Easy. And if the broker doesn't show the charges before buying, change the broker.

Simple! Journal your trades with a special column named charges! Journaling not only helps to keep data, but will develop patience too for a trader. But new traders wont do it until the market teaches them.

I would not recommend parameter optimization to maximize profits as a method for optimizing costs. This AI hint leads directly to overoptimize and out-of-sample will not save you. First cost optimization is in executions since the biggest costs are commissions and slippages -> use limit orders, if order is huge split it with some TWAP o VWAP executions. Maybe the strategy is not good for your current cost tier. Secondly thinking about reduction. Short trades may be much costly than longs, and if their addition compared to costs is not significant, get rid of them. In a backtest, you will usually have the greatest returns with small volatile assets, but in reality they will create you biggest costs and randomness. Be aware!

Not all options are liquid, very few have bid/asks only pennies apart. Most are wider I assume due to lack of volume & open interest. This is something to definitely keep in mind when building strategies using options or other thinly traded derivatives.

In Indian context as on today Oct 2023), assuming it to be one of the widely used and a well known discounted broker, I would simply add ₹30 per side (buy or sell) of the trade on an average and calculate. The trading cost is significant for intraday traders and scalpers but shouldn't worry much for a positional or a swing trader.