Search
× Search
Saturday, May 18, 2024

Archived Discussions

Recent member discussions

The Algorithmic Traders' Association prides itself on providing a forum for the publication and dissemination of its members' white papers, research, reflections, works in progress, and other contributions. Please Note that archive searches and some of our members' publications are reserved for members only, so please log in or sign up to gain the most from our members' contributions.

Chosen technology for Back-testing?

photo

 Adel Humayun, Fixed Income Trader at Daiwa Capital Markets Europe Ltd

 Saturday, November 8, 2014

Guys I am using Python as the weapon of choice for back testing. Have used R as well but finding very useful. Any other technologies that you guys can recommend?


Print

5 comments on article "Chosen technology for Back-testing?"

photo

 Jim Witkam, Owner Altreva, agent-based forecasting models

 Tuesday, November 11, 2014



I like what Gaurav said "...you'd better like to simulate the price series with a certain (or many) modeled process and see how your strategy performs...".

I don't hear this nearly enough but it reminds me of what I'm doing, although I go even further: the strategy is not fixed but adapts/evolves while trading a simulated price series. In fact, the price series comes from a simulated market that is formed at the same time by thousands of trading strategies. So not just one strategy but thousands are evolving and competing against each other. All this is used to generate one-step-ahead forecasts for a real price series.


photo

 Gaurav Singh, HFT Consultant at alphaticks.com (looking for full-time)

 Tuesday, November 11, 2014



Thanks Jim. Appreciate that. I totally agree that unless one competes in the market, she/he can never know the impact of her's/his actions.

I also strongly believe that no matter how much backtesting we may do, its certainly not going to be any closer to the actual performance in the market ! Specially, when it comes to scaling the trading algorithm.


photo

 Thomas Tomiczek, Owner at NetTecture & Trade-Robots

 Thursday, November 13, 2014



With all respect, but all this talk about impact on the market is quite off for most algos and traders. If I run a backtest that does cover 5 years and does like 800 to 1000 trades, uses limit to enter and mostly exit and is in a liquid market then I can totally assume you that the impact of my orders is nil. Zero. Zilch. An occasional 1 tick glitch at maximum.

This is totally different when you trade large volumes (especially with a volume ignorant algo) and / or go fast / hft or try to grab the last of an order book. But going to general "you move the market" really does not do reality justice in 99% of the cases.


photo

 Gaurav Singh, HFT Consultant at alphaticks.com (looking for full-time)

 Thursday, November 13, 2014



@Thomas, like I said one can use miniscule order sizes (temporarily) on historical/simulate data to test only the trade logic, but scaling the algorithm is the biggest challenge, which happens in 99% of the case.

In other words, a trader is not just a spectator in the market, he has to compete with all other participants who will judge him by his actions as much he will judge theirs.


photo

 Jim Witkam, Owner Altreva, agent-based forecasting models

 Saturday, November 15, 2014



@Thomas, how can you know that the impact of your trades is nil? I mean, how can you measure it in reality, since you can't compare the market developments with/without your trade?

Please login or register to post comments.

TRADING FUTURES AND OPTIONS INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR ALL INVESTORS
Terms Of UsePrivacy StatementCopyright 2018 Algorithmic Traders Association