Offering auto-trading long-only options system "sys13"

Discussion in 'Trading' started by botpro, Jan 20, 2016.

Thread Status:
Not open for further replies.
  1. dartmus

    dartmus

    Randomly generated data has been the focus of the discussion in this thread since it began. 80% of the members replying to u have been critical of you. And 80% of that criticism stems from *YOUR* use of RNG.

    If u didn't make absurd claims that those of us (like me) were data vendors trying to sell more data... and instead approached everyone's criticism with an open mind then u would have solved 80% of 80% of your potential customer's complaints and could move on to satisfying potential customers next important wants and desires. Good salesmen listen attentively so they are able to understand what their customers want.
     
    #151     Jan 28, 2016
  2. I just re-visited today this thread now because of your prompting. Thanks for that!

    As I mentioned twice recently (now my third time) whether any basic conventional performance data of the system could be supplied for further elaboration. It appears we still have not heard anything yet. Not even a question asking what sort of performance data do you/we need/want for evaluation from a trading vendor. That is very unusual!

    One thing very very curious, extremely: Using a time frame of 1-second making decisions for many trades lasting for maximum from 10 to 20 days would be very very special! Is it practical or feasible?! Not to mention the cost/premium/spread of buying/holding long options is not cheap.

    Filling limit orders with auto-trading long options based on 1-second time frame would be another great challenge! Whether liquid or slow market!

    Perhaps the system had been traded for a period of time, but now available for sale to the public, probably due to too many failed trades in-house recently! That could be a reason for why the resistance of using actual data for testing.

    Backtesting and forward testing together is basically a must for almost All traders on ET, but it was denied by this offer/system! Why?

    Long options trading is not simple, particularly auto-trading, considering spread, IV, delta, strike, etc. etc. There would be more than 3 data sets including such as underlying/ options/ IV to be required for system development.

    I wouldn't think many or any options traders here on ET would accept that randomly generated data is practically good/close enough for trading actual money. Since usually predicting the movements of underlying/options/IV/etc based on actual data would be quite common, that could not be done by theoretically (pseudo)random data at all, imo! Perhaps every trader/practitioner trading options would think the same!

    Hiding behind actual data through selling a system black box would provide at least one good thing to the vendor when trading mid/long term options. If the system could potentially, even not intentionally, generate both buying long Call and Put options simultaneously distributed to almost half and another half clients (assuming they don't know of each other)! This is exactly what I learned on ET from previous comments in the past by other posters!

    Perhaps making some live calls trading SPY options would be the best forward testing that we could see a proof on ET, imo.



    http://www.elitetrader.com/et/index.php?threads/options-price-movements.297442/#post-4237980


     
    Last edited: Jan 29, 2016
    #152     Jan 28, 2016
    dartmus likes this.
  3. botpro

    botpro

    Oh boy, what a silly question. Of course does the program not exploit the RNG or any such tricks like peeking into the future data.
     
    #153     Jan 29, 2016
  4. botpro

    botpro

    This system does not make use of the Greeks so far, but this can change in the future to further improve it, if it helps.
     
    #154     Jan 29, 2016
  5. botpro

    botpro

    FYI:

    Only seriously interessted people in this offer should participate in this thread. Thx.

    There are currently two interessted parties in this offer.

    You can ask questions also by contacting me via PM.
     
    Last edited: Jan 29, 2016
    #155     Jan 29, 2016
  6. newwurldmn

    newwurldmn

    Only a sleazy vendor would seek to shut down constructive questions.
     
    #156     Jan 29, 2016
    Chubbly likes this.
  7. dartmus

    dartmus

    It is a fact u can't exploit the algos which generate the real data by developing a strategy tuned to RNG data. If you're exploiting anything it has to b the algo which created the RNG data. That exploit isn't transferrable to real data.
     
    #157     Jan 29, 2016
  8. hoppla

    hoppla

    Maybe this has been addressed elsewhere as I did not read the entire thread, but the problem you have here is that you are using a constant volatility in your GBM and price your options accordingly with Black Scholes - how else would you get option prices for it? If you don't somehow model the smile dynamics (maybe you do but I have not read any of this sort) your pricing will be completely wrong and I dont mean by just a few ticks here and there. Also how do you model fills? Are you assuming passive fills only or market orders? In either case you will have slippage.

    Lastly assuming all markets are uncorrelated is too far removed from reality. Here is what I would do so this gets a bit more realistic whilst keeping your cost in check:

    Buy some volatility index bar data like - VIX for example. Then buy some bar data for your underliers. Both can be bought quite cheaply. Use VIX as a proxy for your Black Scholes vol and the underlyings' bar data to price. To determine an estimate of the implied vol level of each underlier maybe build a model using underliers' historical vol and VIX so that each bar you have a mapping from VIX to underlier implied vol - a simple regression may do the trick. Maybe account for the smile a little - since the vol index is just weighted vols over some strikes, assume a realistic, functional form of your smile and generate multiple equity curves by "resampling" the smile so as to skew and tilt it a little in each run. Lastly assume slippage in your executions. You will be surprised where your figures go!

    Won't be perfect but better than what you got at the moment.
     
    Last edited: Jan 29, 2016
    #158     Jan 29, 2016
  9. botpro

    botpro

    The system works with the actually observed volatility, ie. it after each bar generated does update the observed volatility.

    The system uses limit orders only using the last market price, and cancels the order after 45 seconds.
    The system does not depend on immediate fill/execution.

    Ok, that's an interesting argument. I'll see what I can do here. Thx.
     
    #159     Jan 29, 2016
  10. Q
    http://www.option-price.com/

    Option Parameters Call Option Put Option

    Underlying Price 100 Theoretical Price 3.019 2.691
    Exercise Price 100 Delta 0.533 -0.467
    Days Until Expiration 30 Gamma 0.055 0.055
    Interest Rates 5 Gamma 1% 0.006 0.006
    Dividend Yield 1 Vega 0.114 0.114
    Volatility 25 Theta -0.054 -0.041

    UQ
     
    #160     Jan 29, 2016
Thread Status:
Not open for further replies.