146 reputation
3
bio website geektrader.me
location Singapore
age 30
visits member for 1 year, 11 months
seen Aug 30 at 7:30

I am self-trained programmer and trader. Although Aerospace engineer by qualification, I started my career in banking technology and soon moved to Derivatives Sales for a big multinational Investment Bank.

I am currently full time trader at a startup trading firm and use R as primary tool for research.

I am interested in systematic rule based trading and spend most of the time in development and evaluation of systematic trading strategies.


Nov
6
answered Risks of issuing an Autocallable Note
Nov
4
awarded  Autobiographer
Oct
30
awarded  Teacher
Oct
29
answered Calculating Greeks in Covered Calls?
Sep
5
answered ETF Negative Roll Yield
Mar
20
comment Why the implied volatilities calculated are so different
Implied vols for equities show skew. There is nothing wrong with IV being so different for different strikes. Having said that there might be something definitely wrong with your code. Hence my comment instead of answer
Jan
28
comment Generate tick data from candlestick
@DarrenCook Thanks for pointing out "not found" page. It's a new blog I intend to start soon.
Jan
28
comment Generate tick data from candlestick
@DarrenCook I agree with you, I would go on to say that even bar-level backtesting always comes with lot of caveats. E.g. if your stops are typically smaller than bar range, then your backtesting engine has no way of knowing whether you would have got out or not.) Another caveats is the limit orders. OHLC data doesn't have bid-ask spread info nor volume traded at particular price info. IMHO these factors are lot more important than simulation of tick data. FWIW, your answer seemed apt for the question but the purpose of question itself seems bit questionable.
Jan
28
comment Generate tick data from candlestick
What if you create 2 sets of data, first set with each bar converted to 4 ticks with ticks in order of O H L C and the other in order of O L H C and then do your backtesting on both of these data and take worse output as conservative estimate? Just a thought
Jan
28
awarded  Supporter