Dow options system

Discussion in 'Options' started by dottom, Oct 19, 2002.

  1. dottom

    dottom

    I have a 100% mechanical system that can predict a move at least of 1.2% in the DJIA with 78% accuracy (this currently equates to about a 100-pt move with Dow at 8322).

    All back-testing caveats were taken with regards to ensure no curve-fitting. Performance on multiple out-of-sample sets have similar results, all in 76-80% range. The samples-to-input-parameters ratio protects against curve-fitting. Random noise testing was done with input data, as well as monte carlo optimization to observe the 3D parameter surface map to ensure similar results across small changes in parameter input. In short, I am very confident in the robustness and statistical significance of the results.

    So my question is- how would you exploit this? On days that the model says "there is a 78% chance that the DJIA will move at least 1.2% today" what would you do?

    Note that you do not know which direction the move will be, only that the magnitude will be at least 1.2% (the average magnitude is about 1.5% (currently about 124 pts in Dow). 50/50 tendency for up vs down move.

    Would you purchase a Dow futures options straddle? What types of straddles? Would you look at correlated indexes like OEX instead? (1.2% approx results in 5.00-pt move in OEX.) What deltas will you target? Keep in mind there is a bid/ask spread to overcome as well. What combinations would you use, if any. Would you try to go long delta with both futures and options? Keep in mind we're talking a 1.2% move here (the problem would be much simplier if the move was, say 2% or more).

    Here's an example. With DJIA at 8322, you could purchase Nov DJ futures options:

       8500 Call - 2000
       8100 Put - 2200

    If volatility remains constant, if the next trading day DJIA moves 1.2%, one option will likely gain 400 while the other option will lose 300. But will this 100 gain overcome the bid/ask, and also the 22% of the time the move is less than 1.2%.

    Also note that if the market does not move 1.2%, you lose a day's worth of theta and perhaps some gamma. All new positions would have to be entered at close of trading day and exited on close of following day.

    All thoughts welcome. Depending on how this thread turns out I may reveal more of my model. I would like to focus on options strategies to exploit this 78% occurrence rather than on the model itself (for now).
     
  2. Rather than analyzing DJIA and buying options on a "correlated" instrument, try your strategy directly on the OEX since the options are more liquid. If you want to hold only for 1 or 2 days, the theta decay is minimal. If you were not so sure of the magnitude I would say buy a straddle, since it has higher % to be profitable on a smaller move. But since you are very sure of the magnitude, I would buy OTM options at 200 pts (150%+ of the expected move) above and below the market. Basically you want to find the options with the highest ROI. Instead of looking at the absolute strike prices, look at them as P+25, P+50, etc. where P is the market price. So dow is 8000, you expect a move to 8125. You can estimate that at 8100 the 8000 call will go for what the 7900 call is going for NOW, maybe plus some goose factor (volatility). So examine the current market prices and find the options with the best ROI at your expected target price. I would guess that strike prices above (PAST) your target would yield the best results. In fact I would guess that the further out you go, the better your ROI would be, but also the spread gets exponentially worse. So it is a balancing act between the spread and the strike price. (Execution risk vs delta).

    In case it's not obvious, you would have to look at the long side (calls) and short side (puts) separately, and open positions for each side.

    One other idea is to put on a ratio backspread if you are willing to favor one side, and you can get a credit or small debit for doing so. Sell a call ATM and buy 2 OTM calls. You will keep the credit if the stock (underlying) goes down, and make some cash if the stock goes up. In your specific case however, the strikes may be too close for a ratio spread. I estimate about 5 OEX pts.

    Actually come to think of it, you are looking at such a small move (if using the OEX) the only thing you may be able to profit with is a straight-up straddle.

    Oh one other thing.. you could buy futures and hedge with options (I think that was your original thought!). e.g. buy futures and buy puts (favoring upside), or sell futures and buy calls (favoring downside). Since you are going for such a small move, buy the nearest ATM options (or slightly OTM).

    edit: use S&P futures and OEX or SPX options (NOT futures options) for best liquidity.
     
  3. dottom

    dottom

    Assume the model works just the same on OEX as DJIA (it's very close, but I first used DJIA because the parameters used for model input have highest correlation with DJIA - but OEX is so closely correlated I can use model as is and be happy with the correlation. I will modify the model to run specifically on OEX so we have some hardcore numbers to play with, but for now assume 78% chance of predicting 1.2% OEX move).

    So assume 1.2% gain in OEX, about 5.00-pts. Not a lot. Paper trading it I usually target deltas in the 0.30-0.35 range and seem to do okay. If I split the bid/ask spread I am usually lucky to break-even. The bulk of the profits comes from the days that the move is much greater than 1.2%. I could, in theory, look to break-even on days there is only get a 1.2% move, and rely on the large moves for the bulk of the profits.

    I'm hoping some of you options veterans have have ideas on how to extract profits on every 1.2% move! I've looked at a variety of combinations and ratio backspreads but with only a 1.2% move it looks like straight straddle works best.

    Most of the time it looks like all my profits evaporate with the spread. Look at these spreads from today's quotes, OEX at 449.02:

       Nov OEX 460 Call: 10.80x11.50 (delta 0.36)
       Nov OEX 465 Call: 8.90x9.60 (delta 0.30)
       Nov OEX 440 Put: 13.50x14.00 (delta 0.35)
       Nov OEX 435 Put: 11.50x12.50 (delta 0.28)

    Note these are as quoted at close... the actual spread is often less and like I said you can always place a limit order and usually split the quoted spread. But still, you pay on the way in and out. The spread takes away most of my long delta gains on a mere 1.2% move.

    Any thoughts on how a retail trader with a model predicting 78% chance of a minimum 1.2% move (avg. 1.5%) can exploit?
     
  4. I wonder why do you want to bother with options? Execution and reporting is not so good(volume is low at times). You should just determine daily trend and take trades in direction of it. I do not know reasoning behind your system so I cannot be more specific. If you let us know what is based on (momentum, cycles ,break out ) it might be easier to help.
    Walther
     
  5. def

    def Interactive Brokers

    A rep from the CBOE was in our HK office on Friday telling me that the DIA and DJX options are trading something like 80K a day. I didn't have time to confirm those figures but he insisted that the spreads were good and the market was liquid. I've no clue how often the raes/auto-ex system is on line for those products but if it is, it might be worth looking at.
     
  6. nugya

    nugya

    "Any thoughts on how a retail trader with a model predicting 78% chance of a minimum 1.2% move (avg. 1.5%) can exploit?"

    In order to make any money on your model, it needs to predict this move before the markets close previous day.

    You will find that futures trading will probably indicate some movement on underlying(DJI) and the options will be priced accordingly on open.

    My personal experience is that only way to make money on options is when you take the opposite site of the trend or expectation.

    So if you are expecting this 1.2% move you are probably better of selling straddle rather then buying it.

    I would also NOT recommend DJI options as there seem to be little retail activity on them

    GOOD LUCK
     
  7. dottom

    dottom

    One of the reasons the model has such a high success rate at predicting a 1.2% move, is that it is directionally neutral. The model looks at 100% mechanical variables, including volume, volatility, and some custom derivations of momentum & breadth indicators.

    The reason to choose options is that is the best vehicle to go long delta when you can predict the market will move. It's obvious if I said I could predict a 2.4% move (10-pt OEX move or 200-pt Dow) 78% of the time I'd just load up on NTM staddles every chance I get. But the market being as it is, we have to work for our money! :) So the best I could come up with while maintaining a high % success rate is predicting a 1.2% move. I just don't know which direction!

    At this point I'd rather not disclose the model itself, for obvious reasons, but focus on whether a 78% chance at predicting a 1.2% move is exploitable? I'm working on a way to create an 'aggregate indicator' that can't be reversed engineer and may post signals here to further discussion, but again this assumes that you can make consistent profits from a 1.2% move.
     
  8. dottom

    dottom

    Yes, it does exactly that. The model enters on the close the previous day. By 3:45pm I have all the 100% mechanical data needed to make a prediction.

    So as of 3:45pm on Friday I would enter a position and exit Monday at close. Note that if I get a signal on consecutive days, I would just rebalance my long deltas rather than completely exit my initial position and enter a new one (fewer trades = less spread given to the MM's).
     
  9. nugya

    nugya

    If you have a quick look at DJI's recent performance, you will see that 1.2 % daily move is not as rare as one thinks.

    i think, that will be the handicap of your model.You need to test your model on different markets where the % of moves are not as big as DJI's.
     
  10. dottom

    dottom

    The model works for past 6 years. I haven't tested prior to 6 years because one of the inputs I use wasn't available at the time. The model has done very well past 3 months, actually hitting at a 87% rate. (Your observation of frequent 1.2% moves is accurate and results seem to reflect that.)

    Similar results for OEX and SP500 indexes.

    The questions remains- if you were told "the market will move at least 1.2% today" but you don't know which direction, what strategy would you use to exploit this? Or is a 1.2% move not enough to overcome inherent costs in trading (slippage, spread, commissions)?
     
    #10     Oct 19, 2002