Perhaps after clsoing each trade, according to the strike and filled price, Jeff can provide the daily OHLC of the relevant trade-days for the closed trade for our reference. Q Lets take a real example of the data and program outs that resulted in the 10/3/2012 winning call trade: (this data is from the evening of 10/2/2012 to determine a trade for 10/3/2012) SIG DATE..OPTION.....PREV...OPEN..HIGH..LOW...CLOSE 10-02 OCT143C..2.58...2.89...3.00...2.24....2.59 2012..OCT146P...2.57...2.20...2.85...2.05....2.22 ..........OCT144C..1.94...2.27...2.36...1.62....2.02 ..........OCT145P...2.02...1.70...2.26...1.57...1.73 TREND COUNTER: CALL SEQ 2 3 DAY TREND..CALL=13-26-4=L2......PUT=19-12-30=L4 3 DAY DECISION OCT144 CALL: BUY LIMIT=1.74..SELL LIMIT=2.24..STOP=1.21 ------------------------------------------------------------- UQ
One more thing I don't really understand well. Now you observed Just-In-Time that "Program indicates a change of direction." Looks like the 1.85 were filled twice (call and put) for SAR purpose. That means the above two trades don't appear behaving like a general mechanical trading as enjoyable mentioned below. The risk is there could be many times of SAR signals during a short period of time, potentially causing uncontrollable losses. I just try to guess the above, as I know I must be pretty wrong somewhere in my logics.
Your right, the SPY Dec 145 call strike did move below the original Stop of 1.30 on 11-05-12: DATE.........OPTION....PREV..OPEN..HIGH..LOW...CLOSE 11-02-12...DEC145C..1.92...2.15...2.16...1.34....1.34 11-05-12...DEC145C..1.34...1.35...1.68...1.27....1.59 11-06-12...DEC145C..1.59...1.67...2.12...1.65....1.97 But on the first evening (11-02-12) of holding this trade, the program generated a new call signal for trade date 11-05-12. When a new confirming signal occurs while holding an older trade in the same direction, the trading system program mandates a Buy No. 2 with new trade parameters (buy limit, sell limit, stop price) to accommodate the Buy No. 2 price. That means the Stop is moved lower to give Buy No. 2 some breathing room to play out. Its usually moved a minimum of -.15 under Buy No. 2, to as deep as -40% under Buy No. 1 when the VIX is higher (dependent upon the VIX component value input to the program). On the trade example above, the Stop was moved to 1.20 to accommodate Buy No. 2 at 1.35. What would be the point of leaving the Stop at 1.30 with a Buy No. 2 added at 1.35? There would be no breathing room for Buy No. 2 to work itself out. If you read my older posts in the original thread called "My option trades" by ryanpatrick, you will see an extensive use of trades with Buy No. 1 / Buy No. 2 scattered intermittantly from pages 86 to 289. http://www.elitetrader.com/vb/showthread.php?threadid=231231&perpage=6&pagenumber=86 Jeff
Oddtrader, Ok any future trades I enter, I will give the SPY etf value at the exact point of entry and each evenings SPY option data, as well as the mention of new confirming or opposing signals. Jeff
Oddtrader, The trading program is looking at 3 day micro-trends and 9 day longer term trends. It evaluates both of these trend configurations in a comparator program that is looking at thousands of signals of the same trend configurations that have occurred over many years, and it makes a decision to: (1)buy calls....(2)buy puts...(3)buy call & put....(4)..No Trade. It has done this with 70-75% accuracy for several years. So on the example your talking about: Program indicates a change of direction. Closed open trade: SPY NOV 143 puts at 1.85 (breakeven). opened new call trade: SPY DEC 145 Calls Buy Limit: 1.85 (filled) Sell Limit: 2.40 Stop: 1.30 The SPY Nov 143 put trade above was generated from the program on the evening of 10-31-12, for a trade on 11-01-12. The SPY Dec 145 call trade was generated from the program on the evening of 11-01-12, for a trade on 11-02-12. Entries for the put and call trades were on different days. The trading program generates new signals almost every evening, and sometimes in both directions on the same day. Risk is limited to the drop from the entry price to the stop, and that is a variable dependent upon: 1: Is there a Buy No. 2 that requires deepening the Stop to give Buy No. 2 a chance to work out? 2: Is the VIX high? requiring the deepest maximum Stop of -40% under Buy No. 1? Sometimes on very rare occassions the market will open severely opposite my option trade direction and I will Stop below my planned Stop value. It happens but its so rare its not worth considering. One last thing: Unless a trade is 4 days old, the trading program doesn't tell me to close an open trade in one direction because there is a new trade in the opposite direction (like the SPY Nov 143 put and SPY Dec 145 call trades above that you mentioned). That was a personal decision, not the program's decision. Who was smarter? The program of course! because had the put trade above been left undisturbed it would have easily won the next day. Same with last month's (October) put trade that I closed it about 1 hour after I opened it because I had to drive to Washington for an emergency. Had I left it alone and let the bracket order work, it would have won the same day. My personal decision's when I interfere with the trading program are about 40-50% accurate at best. The trading program is 70-75% accurate over several years. Jeff
Thanks for your feedback Jeff. Just curious: How do you know/determine whether next time you should go fishing comfortably without worrying when this time you decided not to keep/leave a trade open when driving to Washington? Q Quote from jeffalvinson: At this point I have nothing to do with the final trade decision. Its up to the final decision maker (Market Monitor Program). I simply enter the trade with semi-automated orders and go fishing....or to the movies....or clean the grease off the driveway. "Its the machines I tell you, they are taking over." UQ
Thanks. That would be nice! Where do you get your data services for backtesting with options daily OHLC?
The SPX (SP500) is the only major index that is still clinging at or near the 200 day moving average (see attached chart). The Nasdaq (COMP), Dow Jones(INDU), QQQ etf, Russell (RUT), IWM, & IYT have all moved well below the 200dma. The only other odd thing is the financial etf the XLF is still well above the 200dma. Is this because the Fed is buying $40 billion a month of the crap the banks have on their books for an indefinite amount of time? The question is: "will the SPX & XLF lead us higher from here, or are they just followers?" Questions like the ones above are why I just trade like a mechanical monkey.
But there is a serious fundamental flaw in backtesting using the Open, High, Low, & Close with historical data. Here it is: Real Example from Friday 10/26/12 on SPY Call option: Option.......Open.....High.....Low......Close NOV142C...1.86......2.04.....1.49..... 1.83 (all exchange data only records those prices) Lets say the above data was older data and your testing your new trade system for a call trade on 10/26/12 with the following trade parameters: Buy Limit: 1.55......Sell Limit: 1.95 (+25%).....Stop: 1.15 (-25%) Looking at the old data above: You have no way of knowing if the "Low Price"occurred before the "High Price" ???? Without knowing that piece of intra-day information you cannot assume the low occurred before the high, so you have no choice other than to record your test trade as filled (Buy Limit: 1.55) "but holding overnight." Now the next day after holding this call option the markets could have gone real negative....took out your Stop: 1.15 (-25%) and you had to record your back-tested call trade as a Loser! The "reality" of what really happened on Friday 10/26/12 was this: Option Opened at 1.86....and then tanked to 1.49.....and then rallied to 2.04.....and then pulled back and closed at 1.83. So your test trade was actually a winner but without intra-day data you are forced to record it as a loser. As far as I know in the last 15 years, all exchanges only record OHLC. This type of inaccuracy can effect a back-tested system's results by a rediculous amount. The best thing to do is simply test a new system forward for about 1 year before drawing conclusions. Jeff