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

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

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

    botpro

    Hi everybody,
    I've over the past 6+ months developed and tested a long-only-options trading system
    that I would like to offer interessted firms and individuals.
    Below please find the loose 'prospectus' of the system along with some test data.
    Please feel free to PM me for any questions you might have.



    Prospectus of the auto-trading long-only options trading black-box system named "sys13"
    Author of the system: U.M. in Germany (PM botpro @ ET)

    - Intended audience: trading firms, hedge funds, banks, professionals and also retail traders with big account

    - System is intended for automatic options trading (mainly equity and index options).
    The system trades long options only (ie. long calls and long puts only; does not use options shorting/selling).
    It needs multiple underlyings (stocks, indices) for trading their options, the more the better.
    It could also work with other options, like futures options.

    - Since everything needs to have an identification: the name of the system is simply "sys13" :)

    - The system needs to trade multiple instruments (ie. options) at the same time by multiple bots,
    There should be at least 5 bots, better is 10+. And it needs at least 1/3 more tradable instruments
    than there are bots; the more the better to choose from, especially for skipping or unloading flat instruments.
    These limits are because the capital has to be traded by multiple bots, as said,
    each bot trading only a single instrument. So it gets harder to trade more than about 50 instruments
    at the same time because we are interested primarily in the most active and liquid instruments.
    The equity curve gets smoother (ie. less drawdowns) the more bots are trading the account.
    From that it follows that the inital trading capital should be at least 100k and at most about 600k.
    Of course the upper limit can be increased by trading also at some foreign exchanges (ie. Europe, Asia etc.).
    An example:
    10 bots (= 10 different instruments), each starting with 15k --> 150k invested,
    initial account size should be about 4x of that because the system needs to
    apply also its key-strategies after opening the position, ie. adjusting/hedging until it reaches its target.

    - This system does not and cannot predict the future, so instead it applies
    some other (proprietary, ie. secret) methods based on the price action and the data of each position.

    - The system is tested using Monte-Carlo with simulated bar data (702 bars/day, ie. 33.33-sec bars @ 23400 tradingseconds/day)
    using GBM (Geom. Brownian Motion) and the Black-Scholes options pricing model, but limiting itself to market rules (s.b.).
    For backtesting (and forwardtesting) the number of bars per day is configurable:
    max. 23400 bars/day is possible (ie. 1-sec bars), but then of course the execution
    of the program gets significantly slower (on a 4 x 2500MHz system).

    - The author is an experienced senior C++ developer, and the sole author and developer of the system, and residing in Germany.
    It took more than 6 months to develop this system. It's written in C++, uses STL and some goodies from the latest language standard C++11.
    It uses some new and novel trading ideas and algorithms. Because of that, it is a black-box system,
    ie. the source code will not be sold, nor disclosed; it is a trading secret, ie. the "edge" of the system.

    - Platform: Linux. If needed, can be ported to a Linux-like environment like Cygwin under Windows.
    Ie. it needs a recent C++ compiler that understands C++11, preferably GNU g++ v4.9 or higher.

    - The system is intended primarily for the US options market (because of the high volume
    and number of tradable instruments, and the availability of APIs by the US-brokers),
    but can also be used with foreign brokers/exchanges/bourses as long as an API is available for auto-trading.

    - It is not an HFT system, so commission is not an headache (avg daily round-trip-trades is about 16 when 50 bots trade).
    It is also not a pure daytrading system as the target of each bot needs on average up to 5 days to be reached,
    so it can be said it is a swing trading system that is fulltime connected to the exchange.
    It needs to know only the current price and quote (ie. Level-I),
    but it does not need any deep-going orderbook data (ie. Level-II).

    - It does not use any known TA indicator or oscillator, it rather uses some new ideas
    and algorithms based on statistics and probability.

    - It adds autom. more bots if possible, so that more of the available cash gets traded.
    Of course there are some realistic upper limits to obey of what is tradable and what not.

    - It limits itself by not trading more capital than is available at day start,
    ie. it follows the settlement requirements in conjunction with margin account.
    Otherwise even more profit is possible.

    - The daily invested capital is on average less than half of the account value.
    Otherwise even more profit is possible.

    - It works in every market situation by autom. adapting to the new situation.
    It tries to detect and avoid flat markets as this is poison esp. for options.

    - Performance: since this is an options trading system the performance can be quite very high.
    Of course it also depends on the parameters (ie. volatility, maturity/expiration, strike (ATM, OTM), etc.).
    +1000% p.a. or even more is very well possible (this depends also highly on the initial account size).
    A warranty of at least +100% p.a. can be given.
    --> cf. test data and graphs in the examples.
    Code:
    Summary of 5 runs (each 252 trading days, ie. a trading year), details s. Examples:
                               |     1    |     2    |     3    |     4    |     5    |
    ---------------------------|----------|----------|----------|----------|----------|
    Profit in 1 year           |  962.53% | 1173.98% | 1101.22% | 1042.17% |  968.07% |
    Avg Monthly Rate of Return |   21.77% |   23.62% |   23.02% |   22.50% |   21.82% |
    Avg Weekly Rate of Return  |    4.80% |    5.18% |    5.06% |    4.95% |    4.81% |
    Avg Daily Rate of Return   |    0.94% |    1.01% |    0.99% |    0.97% |    0.94% |
    Highest Daily Gain of AV   |   14.11% |    9.56% |    8.35% |    7.99% |   10.42% |
    Highest Daily Loss of AV   |   -5.05% |   -2.79% |   -3.06% |   -4.09% |   -3.55% |
    Max Drawdown of AV         |   -8.46% |   -3.42% |   -3.06% |   -4.09% |   -7.14% |
    
    - Examples: see sys13_tests_v2.zip
    The examples contain at least 5 sets of annual daily data using a 600k account (as csv and as xls;
    xls has same data as the csv, plus some metrics and charts).
    The daily gains are reused, ie. compounding is implicitly applied.
    Test params: InitialAV=600k, Vola=30%, Strike=ATM, ExpDays=3m, MinInitialPosSize=15k..25k, ...
    Analysis: as can be seen from the InvestedPct-Chart: adding even more bots beyond the used limit of 50
    would bring even more profit (would require an at least twice faster PC, ie. 8-vCore or more @ >=2500 MHz)

    - Price for leasing the black-box system: 15% to 20% (negotiable) of the profits the system makes.
    Contract/License: 1 or 2 years, renewable

    - The system needs to be integrated into the API of the broker (estimating 8 weeks).
    The author can do this work and also the administration, oversight, support.

    - If you need to trade more capital than an initial 600k then more instruments need to be traded
    and a powerful machine or multiple of them need to be used. We can find the right solution.

    - What you get
    1) a very good black-box options trading system that autom. adapts itself to all market situations
    2) the author and developer of the system: a professional senior C++ software engineer
    and non-professional researcher (trading, maths, crypto, sw/hw)
    ie. instant support, improvements, testing and adding new ideas and features etc.

    - Contact: PM botpro at http://www.elitetrader.com/

    .
     
    Last edited: Jan 20, 2016
  2. ktm

    ktm

    1000% a year and you are offering it to lil ole us?

    I for one.... am honored. I thought we were a more raucous bunch with a sullied reputation and no prospects for untold riches. Clearly, I was mistaken.
     
    dratsum likes this.
  3. botpro

    botpro

    The system requires a fair amount of initial capital: the test were done with 600k.
     
  4. botpro

    botpro

    Addendum:
    PM me if you would like to test the system with different parameters (InitialAV, vola%, expiredays, min_bots, max_bots, min_initial_pos_amt, strike_off%, max_invested% etc.)
     
    Last edited: Jan 20, 2016
  5. trilogic

    trilogic

    give it a runs with a 10k account and you may strike up some interest hwre
     
  6. botpro

    botpro

    :) no, such an account size is unfortunately hopelessly too small for this system,
    and don't forget my provision of 20%: in the end it shall be lucrative for both sides, the customer _and_ me...
     
  7. trilogic

    trilogic

    ok, so what would be min allocation ?
     
  8. 2rosy

    2rosy

    Am I reading this correctly? You tested against data that you generated
     
    dratsum likes this.
  9. botpro

    botpro

    The absolute minimum is 100k, as was stated in the initial posting:
    "From that it follows that the inital trading capital should be at least 100k and at most about 600k.
    Of course the upper limit can be increased by trading also at some foreign exchanges (ie. Europe, Asia etc.)."

    The result of the system depends on such main parameters, in other words: the result will be less with 100k, and the curve not that smooth (ie. drawdown could be higher).
    I'll shortly post the result of this setup...

    Ok, here's it. As was feared by me, with small account sizes the drawdown gets too much :-(
    Code:
    Profit in 1 year              845.96%
    Avg Monthly Rate of Return     20.59%
    Avg Weekly Rate of Return       4.56%
    Avg Daily Rate of Return        0.90%
    Highest Daily Gain of AV       38.33%
    Highest Daily Loss of AV       -9.62%
    Max Drawdown of AV            -29.45%
    
     
    Last edited: Jan 20, 2016
  10. botpro

    botpro

    Yes, that's correct: an algorithm called 'Geom. Brownian Motion' was used for generating the data.
    More info here: https://en.wikipedia.org/wiki/Geometric_Brownian_motion
    "A geometric Brownian motion (GBM) (also known as exponential Brownian motion) is a continuous-time stochastic process in which the logarithm of the randomly varying quantity follows a Brownian motion (also called a Wiener process) with drift.[1] It is an important example of stochastic processes satisfying a stochastic differential equation (SDE); in particular, it is used in mathematical finance to model stock prices in the BlackÔÇôScholes model."

    In my test environment each run generates different data.
    Program works bar-weise: creates a single datapoint (bar) and that gets sent to the trading bot(s), they process it (ie. open/close/adjust the pos), then the loop repeats...
     
    Last edited: Jan 20, 2016
    #10     Jan 20, 2016
    marketsurfer likes this.
Thread Status:
Not open for further replies.