Thread by @Ankit_Quant

No Matter how many algos I hv created in past, evry time I bring a new algo to LIVE trading-I'm jittery. That is the human side of Quant Trading. We build robots/systems – but ourselves are humans with same apprehensions and self-doubt as anyone else. A 🧵of my learnings 👇: 1/n

from bringing #QUICKSINGLES to research stage to LIVE Investor Deployment. It has hit a new ATH today and now has had 60 trades to it on LIVE Trading. Some background of research and approach first: 2/n

I had bt this strategy and associated robustness testing was completed in Mar-21. For the first time I was experimenting with a pure Data Driven strategy research (it had no hypothesis when I started my research). I was skeptical of my BT and kept on finding ways to break it. 3/n

The backtest was on Index, however I wanted to use it on buffer capital lying idle on my Trend Following system – DYNAMIC. Thus, I could not trade it on Futures as it would require more margin. I decided to trade it on options. 4/n

My 1st mistk was nt to put correct Sizing Options trade. Foolishly, I used prem of opt as measure of risk and set risk as %age of pf. If an option is 100 Rs and 50 lot size, 1 lot will risk 5000 INR. If I am risking 5% on 3L capital, I traded 3 lots. This was a big mistake. 5/n

Because in a way I created an execution that will trade higher lots (and thus have higher DELTA exposure) close to expiry/low IV situations and will trade lower when away/higher IV. This was a mistake because my system had no view on day of week/IVs. 6/n

By setting up such execution, I let unforeseen risks creep into my execution – even though my BT did not have it. I quickly realized this in first 4-5 trades (the dip in P&L you see above) and adjusted my approach. 7/n

I changed it to constant DELTA exposure and made tweaks on DTE based expiry selection to bring execution as close to BT as possible. Since my holding period was small, I decided to assume the risk of assumption of DELTA changes not materially impacting my performance. 8/n

This experimentation was done only on my LIVE account, with no client being allowed to deploy it LIVE. This whole process took a good 2 months and by May-21 I allowed investors to deploy the algo. 9/n

And as luck may have it, the strategy hit a rough path immediately after that. I was in-money coz of my earlier deployment, but investors were not. We had to wait a good 3 months before hitting the kind of home run we expected it to hit. 10/n

Imp Lessons -
1. Trading is a game of patience. Not once, not twice – but day after day, wk after wk, yr after yr. Good sys and research processes are slow moving objects. And even then you get junk out of it most of the times. 11/n

2. Being paranoid about risk is not a bad trait to have. You can't flush out sys out of a magic hat regularly. Your sys is bound to fail in LIVE trading if you haven't robust tested it. Focus on breaking your BT before you take it live. 12/n

3. Options is another way to express your opinion on markets. If you're not learning Greeks, you are flying a F-16 with your blinders on. You'll crash and burn sooner or later. Do not be fooled by trainers who have no skin in your game. LOVE YOUR MONEY. 13/n

4. Risk can never be completely hedged away. It can only be transformed from one form to another. By trading options I gave up one form of risk and took up another. I made assumptions that brought other risk. Be cognizant of that. 14/n

Lastly - its ok to fail. Its ok to stumble and get up and try again. That's what makes this field so exciting. Good luck with your trading. END n/n