No Positions – October 12, 2011

[Billy]

Forum Clusters 111012.xlsx

Conflicting ST/LT statistics for the IWM robot and neutral edges for the GDX robot do keep us safely in cash, waiting for the next objectively reliable setups and signals.
Here are the updated charts.
Billy

[adam ali]

Billy,

I'd like your take on something. In reviewing the action of $TICK in the last hour yesterday, it was positively sloped even though the SPY essentially traded flat during this time period. Would you treat this as a positive divergence in looking at how the next day's trading might play out, or does everything sort of "stop" when the particular trading day is over and $TICK analyses start afresh with the next day's trading? Is there a carryover effect?

[Harry]

If Billy hasn't answered, I'd like at add a thought to this question.

In addition to day-to-day carryover, what about same day impacts? Recently, I read a article quoting Doug Kass attributing much of the recent volatility due to ETF's and leveraged ETF's as they need to re-balance their holdings by the close. I don't know enough to comment if this would be a large enough impact to move the markets? Anyway, this article and Paul's recent TICK comments on the extreme market moves in the last hour or so has prompted an idea - are the Algo's getting turned on late in the day for these ETF's and can we anticipate the moves by watching the underlying index up to a certain point during the trading day (i.e. 3 PM)?

I do not have the ability to backtest such an observation yet, but I wonder what the winning percentages are?

By the way, I assume the answer is no ... but wanted to solicit Billy's thoughts before I ruled it out. I know we can get many nasty reversals in the last half hour of the trading day.

[Billy]

If Billy hasn't answered, I'd like at add a thought to this question.

In addition to day-to-day carryover, what about same day impacts? Recently, I read a article quoting Doug Kass attributing much of the recent volatility due to ETF's and leveraged ETF's as they need to re-balance their holdings by the close. I don't know enough to comment if this would be a large enough impact to move the markets? Anyway, this article and Paul's recent TICK comments on the extreme market moves in the last hour or so has prompted an idea - are the Algo's getting turned on late in the day for these ETF's and can we anticipate the moves by watching the underlying index up to a certain point during the trading day (i.e. 3 PM)?

I do not have the ability to backtest such an observation yet, but I wonder what the winning percentages are?

By the way, I assume the answer is no ... but wanted to solicit Billy's thoughts before I ruled it out. I know we can get many nasty reversals in the last half hour of the trading day.

Adam and Harry,

1) ETF’s holdings rebalancing must be executed by the close but are actually spread and averaged all over the day by market makers through constant arbitraging. So this has minimal to no impact on the final hour’s $TICK readings.

2) From a discretionary point of view, it has become useless to make intraday decisions (for position and swing trading) before the very last minutes of trading. This is true whether using $TICK, 20 DMF or VWAP. Last hour’s most powerful algorithms have constantly triggered and annihilated any conclusions that could be made from the earlier day’s action.

3) I read cum TICK as a market makers algo detector and 20 DMF as a large players algo detector. I want to see them behaving in synch to feel comfortable with holding positions overnight. I give much more weight to the 20 DMF. For example, a strong last 30-minutes 20 DMF with a weak cum TICK suggests that the cum TICK drop is just a very short term tactic by market makers to re-load their own inventories because they are seeing large players accumulation growing. They like to reload around the day’s VWAP and around pivot and other floor levels, hence the help from the multi-pivots roadmap.

4) For the same reason, yes, it is important to take into account cum TICK divergences with price like the one on Tuesday mentioned by Adam. All we saw were rising cum TICK moving averages and cum TICK holding above the averages and bouncing above the ½ day average. We could discretionarily conclude that the uptrend was not over yet. A close below any of the averages would have been a warning sign.

5) As soon as the automatic real-time 20 DMF will become operational, Pascal will backtest it with cum TICK strategies. If significant edges can be found, this could become the foundation for a real-time robot tracking the highest probabilities of success from the combination of large institutional players programs (20 DMF) and market makers programs (cum TICK). Other strategies with high probabilities of success would also be included.

Billy

[Adriano]

Adam and Harry,

3) I read cum TICK as a market makers algo detector and 20 DMF as a large players algo detector. I want to see them behaving in synch to feel comfortable with holding positions overnight. I give much more weight to the 20 DMF. For example, a strong last 30-minutes 20 DMF with a weak cum TICK suggests that the cum TICK drop is just a very short term tactic by market makers to re-load their own inventories because they are seeing large players accumulation growing. They like to reload around the day’s VWAP and around pivot and other floor levels, hence the help from the multi-pivots roadmap.

Very interesting, thank you.

4) For the same reason, yes, it is important to take into account cum TICK divergences with price like the one on Tuesday mentioned by Adam. All we saw were rising cum TICK moving averages and cum TICK holding above the averages and bouncing above the ½ day average. We could discretionarily conclude that the uptrend was not over yet. A close below any of the averages would have been a warning sign.

I think yesterday was also the case.

5) As soon as the automatic real-time 20 DMF will become operational, Pascal will backtest it with cum TICK strategies. If significant edges can be found, this could become the foundation for a real-time robot tracking the highest probabilities of success from the combination of large institutional players programs (20 DMF) and market makers programs (cum TICK). Other strategies with high probabilities of success would also be included.

Billy

Sounds fantastic. And I'm happy you guys keep having great ideas and work to improve the already excellent robots.

[adam ali]

Excellent explanation, Billy. As always, terrific work done by you, Pascal, Bob and Mike - and all the community members.

[Harry]

Agreed. Thank you for the very informative response! I am continued to be amazed by how much I keep learning day to day.

Harry