Thursday, June 20, 2013

Keybot the Quant Turns Bearish

Keystone's trading algorithm, Keybot the Quant, completed the flip to the short side at the opening bell today. Commodities broke down again with GTX under 4764. Utilities remains weak. XLF is tempting failure at 19.29. As always, stay alert for a whipsaw today or tomorrow, however, this market move to the short side will gain serious steam if XLF loses 19.29. More information is available at Keybot's site;

Keybot the Quant

9 comments:

  1. KS,

    Earlier you mentioned that Keybot would wait 90 minutes to go short, if the market had a gap open of 11 points or higher.

    So, I would expect Keybot to register a short entry price around 1603 on the S&P, not the very theoretical 1625 price. The S&P closed yesterday at 1629 and the ES futures were down16 points at the open. So, how could your short entry be at 1625, just 4 points down from the previous close? And, were you just kidding when you mentioned that Keybot would wait 90 minutes to short if there was a big gap down open?

    ReplyDelete
    Replies
    1. There are many background routines running, the 90-minute timer is important, and it triggered, but is was a moot point since the commodities collapsed due to China overnight, that overrides the other parameters and immediately wanted the algo to click short since utes and commodities have collapsed over the last two days.

      Delete
  2. Keybot could not have shorted 1625. 1613 at most. I was watching at 9:30 AM. Look at SPY. It opened at 161.86 equivalent to SPX 1612.27.

    ReplyDelete
    Replies
    1. Alex, we go through this every time. The SPX number is a means for the program to register a comparison percentage. The SPY number is what is important which is what you have listed, that is correct. The SPX is more at 1616 but again, it is not a major factor. That is why the actual loss is much greater than the program loss.

      Delete
    2. Right, but it's a very misleading statistic to report, so why report it at all?

      What occurred today, as well as many other times, is that after the bell, market skyrockets/tanks. Even though KB algorithm didn't update an hour after the opening bell, it essentially goes back and says "Well, according to the algorithm, it would have shorted at [1625 in this case], even though the market opened lower [1613 in this case]. I mean, hell, wouldn't we all love to have that ability!

      So why even bother reporting that number? I get that your actual trading metric is the thing to focus on, but it gives KB way too much credit. There's a reason people bring this up every time KB goes back and backdates its position.

      I mean I do love the site, I really do. But please try and keep an open mind with this.

      Delete
    3. You are misinterpreting the description Anon. The only time this comes up is on the gap ups and gap downs. Keep it simple and follow the ETF trades only and you will be okay. Think of the SPX number as the purity trade. In an ideal world if there were no timing constraints (seconds and minutes), no commissions or fees, no mechanical and program ineffficiencies, that would be the trade, so it is not misleading at all, think of it as the perfect trade. It is a useful number since it highlights the difference between an ideal flip and the actual flip that results in the actual trading percentage and the number that matters. The program percent accurately reflects the pure program if the world was perfect and you could make a perfect trade each time. The only reason the question surfaces now and then is when the gap up or gap downs occur since the SPX and ETF's will differ at the opening marks, the ETF's will always move more drastically in the new trend direction. Since this happens rarely and amounts to a few handles at most it is of no consequence to the program percent overall, and typically on the return move, the affect lessens even more.

      Delete
  3. KS, that's where you're wrong. The SPX and ETF do not differ by 12 points (1625 vs 1613) because the ETFs get extended because they "move more drastically"... if that was the case, there'd be GROSS amounts of arbitrage.

    Among the reasons they differ is because the SPX number doesn't get updated until the markets open. The ETFs, meanwhile, have been trading for hours pre-market. That's why to focus on the pre-opening SPX number, and then somehow saying that KB could "perfectly trade" that number, is preposterous.

    ReplyDelete
  4. KS,

    Is there any reason why you do not use the SPY instead of the SPX, which is not a tradeable instrument?

    I believe using SPY would add more credibility to your system results.

    You do have an interesting site and a worthy model. But, I agree with the other anon comments above that it is misleading to use the theoritical SPX values, which reflect an instrument that does not trade in the real world.

    ReplyDelete
  5. Whoa, mutiny on the bounty, mutiny on the bounty. LOL There is a lot of misunderstanding of what is occurring. Do not get too wrapped up into the middle SPX program number. That number is a reference. It has no impact on the future turn for the algo. You are giving it much too much emphasis. The benchmark SPX and the actual trading percentage is what matters, this is the money trade. The SPX benchmark changes continuously and is easy enough to calculate, simply compare it to 1426. Any hedge fund or other fund uses the SPX as a base of comparison, that is why it is listed. The actual trading is the number that reflects the actual end result trading, this is the important number. The SPX is simply the underlying index and serves as a reference. For the next turn which will be back to the short side, watch the ETF. The SPX reference can also go either way since the model cascades, everything that comes after is affected by what happens before, so all the moves are double-edged swords that can go either way. Think about it this way. If you are a jogger and you will flip long or short if you see a red bird, and the path is lined with mileposts, so you run along and all of a sudden you see a red bird, you make the change and then at the same time you write down the milepost number as you run along. The milepost is a reference, the bread and butter and important number is the actual trading numbers which is when you saw the bird. The algo is rigid once it changes but for example, tomorrow the SPX may run up to 1640, but if parameters such as RTH and XLF remains bear friendly, down will remain the proper direction and you are actually better to enter a short trade at the higher relative SPX value, which you would benefit from. So, do not pay as much attention to the underlying, the SPX, as you should to the timed changes for the ETF's. For the rare gap up or gap down days like today the affect on the SPX reference is negligible. You must compare the few handles to the long path all year long for the SPX that can cover a few thousand points. The SPX moved about 200 points in just the last few days (count the ups and downs).

    ReplyDelete

Note: Only a member of this blog may post a comment.