## Blog / Our Decision making process about preferred direction for overnight position. #2

### Our Decision making process about preferred direction for overnight position. #2

**Analyzing the Current Prediction Data Puzzle.**

Before we try to solve the predictor puzzle having predictions for the six major indices on three time frames plus the Omega predictions for the NASDAQ / NASDAQ100 and the S&P500, we make some general considerations.

**Index Value Calculations:**

The index value calculations use a so-called CAP – weighted process, which means, that the company which have a bigger capitalization, will have a bigger impact within the index. But just because of the huge cap differences even within the NASDAQ100, the NASDAQ / NASDAQ100 uses a so-called modified cap weighted calculation mechanism, which assures, that the bigger capitalization companies have bigger impact, but their impact is not 10 times bigger, if their capitalizaton is 10 times bigger than the other index component. This way the order of importance within the index is maintained, reserved and determined by the cap size. (Note that the DOW index is a price-weighted index.)

**Index Size:**

The biggest index by far is the S&P500 index, than the NASDAQ index, than the S&P100 index. Smaller is the NASDAQ100. Much smaller indices are the S&P400 (Mid Cap) and S&P600 (Small Cap) indices.

**Index components and interrelationships:**

Note that some companies might be member of multiple indices. For example Intel (INTC is a NASDAQ company, but also member of the S&P500 the NASDAQ100 and the S&P100 indices.) This means that we can see / calculate correlations between indices.

During early 2010, the following overlapping were present in the indices:

**71** NASDAQ100 companies are also member of the S&P500 list of companies (These amounts to **14.2%** of the companies in the S&P500.

**100 or 100%** of the S&P100 index components are also member of the S&P500 index (These amounts to 20% of the S&P500 companies.)

**93** NASDAQ companies are also member of the S&P500 index. (These amounts to **18.6%** of the companies in the S&P500 index.)

**14 or 14%** of the NASDAQ100 companies are member of the S&P100 index. This also shows that technology is a relatively small, though important part of the US economy and the indices reflecting the whole US economy.

Obviously None of the S&P400 (Mid Cap) companies are member of any other major index we calculate.

None of the S&P600 companies are member of any other index we calculate.

So if a huge tech company make a big price move, than the Biggest impact is in the NASDAQ100 index, a bit smaller within the NASDAQ index, and again smaller the impact in the S&P100 and S&P500 indices, assuming that the move in the given stock is not moving much the other index components, it is an isolated move.

Conversely if a big S&P100 componet make a big price move, usually the impact on the index itself is smaller, as the S&P100 and S&P500 index is more diversified than the NASDAQ / NASDAQ100 indices.

Note that these numbers are changing every year, as indices get modified, if a company get out of the index for some reason and a new company get in.

**Analyzing the prediction data:**

We would like to note here that the market is constantly impacted by the cumulative effects of many different cycles (4 Days cycle, 4 weeks cycle 4 years cycle...) and the impacts of that is, that the market changes direction, but not necessarily at the beginning of the week or at the beginning of the month, so those direction changes usually occur in between of those prediction periods.

When we selected the Index and the asset class to play we check first the** Daily predicted direction for that index, and the win probability of that prediction.**

For example we play the NASDAQ index, and the prediction is Bullish, with 77 % Win probability.

That is a nice win probability, as it is a bit above average, but that alone is still not enough to make the decision to open a position to hold overnight.

So we would like to collect more confirmation. We check, if the other major indices also predicted to move in the same direction. If both the NASDAQ and the S&P indices predicted to move in the same direction, than that information mean two things. The win probability is a bit increased (by 1.6% - 3.8%) and that the market more probably makes a bigger move and not hesitate and stay in a small range or struggle to move, if the predicted direction of the major indices shows some divergences.

**Second we check the weekly prediction.** Here we consider three things.

- The predicted weekly direction.

- The predicted weekly win probability.

- The time / date or how old is that weekly prediction is.

If the win probability is about average for the week (Say 78% for the NASDAQ) than it does not add much to the picture, but if the win probability is much bigger, than the average, say 85 – 93%, than it increases the win probability of our overnight position, especially if the prediction is freshly calculated (We calculate that on Friday for the next week, assuming that Friday is a market day.)

If we are already in the second half of the week (Wednesday evening or Thursday evening) than the weekly prediction have minimum impact.

Weekly prediction data might increase / decrease win probability by 2-6% depending on the status of the above three points. (The biggest support being a fresh Weely prediction with high win probability pointing in the same direction as the Daily prediction.)

**Third we check the monthly prediction.** We acknowledge, that this data have minimal impact on the decision for overnight position, even at the beginning of the month, when this prediction is relatively fresh. It has less, than 2% impact on the final win probability. It might have a role for the position size selection.

**What we consider next and really important is the Omega prediction.**

Omega prediction have major impact on our decision, because of many reasons. As we mentioned, it has relatively low correlation with the other daily predictions, and fails rarely at the same time, when the Daily prediction fails. We also stated, that the Daily prediction is better to predict the market movement at the beginning of the prediction period (For Daily prediction namely for the first 30 – 90 minutes.) whereas the Omega prediction is better to predict the market movement for the second half of the day.

The Daily predictions have little capability to predict the direction of the Gap, (GAP prediction capability is roughly 52%) but the Omega prediction has a much better capability to predict the GAP direction. (It was about 65% during the past 6 month till Nov 2010.)

During the previous 6 month period, the average Win probability of the Omega prediction for the NASDAQ was 72%. If we require, that the Omega prediction AND the Daily prediction must point in the same direction to open overnight position, than the statistics will change notably.

During the same period the Win probability for Long positions was 78.2% and the win probability for short positions was 87.23% On average the win probability, if both the Daily prediction and the Omega prediction pointed to the same direction was 82.79% This means, that we increased the Win probability by about 10% from 72 to 82.79, just by adding one condition to the equation.

The other impact of this was, that we had much less occasion, (Less number of trades) when this requirement fulfilled. It happened 54% of the days / cases.

I would even say, that the impact of the Omega prediction is bigger then the impact of the daily prediction, having average Daily prediction win probability. If we have GAPPY markets, when the overnight risk is increased, than the Omega prediction play even bigger role.

We do not have defined 6 different states for the Omega prediction, as we did for the Daily prediction.

(Very Bullish, Bullish, Neutral with Bullish bias, Neutral with Bearish bias, Bearish, very Bearish)

For the Omega prediction we might state only the Following:

Bullish, Bullish to Neutral, Bearish, Bearish to Neutral.

If the Omega prediction is Bearish and the Daily prediction is Bullish, than we do not open overnight position. If the Omega prediction is Bearish to Neutral, and the Daily prediction is Bullish, than at least we do not have clear argument, against opening a Long position. So if „Neutral” mentioned for Omega prediction, than the Daily prediction direction might be played, even if the two prediction not completely agree.

If we play the NASDAQ / NASDAQ100, then the predicted direction for the S&P400 and S&P600 has also minimal relevance. Small and mid cap companies play a role in the breath – type indicators, but have little direct impact on the bigger indices. The impact direction is usually the reverse, so that the majors might pull them up / down even more than they move.

The impact of a 5% prediction win probability difference could be very high for our yearly trading result. Let’s see that in the following example:

Assume we trade the overnight prediction data 100 times a year or about 40% of the total yearly trading days. Since we are taking the risk of the overnight GAP, the Average Win size is close to the average loss size. If we have a 75% win probability, playing with a 100 K account and the average win and the average loss is also 1K, than our yearly profit is 75K – 25K = 50 K $ (Not calculating with the reinvestment of the gain.)

If we manage to increase the Win probability by 5% to 80%, than the total win will be 80K, the total loss will be 20K, so the result is 80 – 20 = 60K, which is a 20% better result. (If we are calculating with the reinvestment of the periodic gains than the difference is even bigger.)

Note however, that the overnight risk is something we need to take very seriously. The overnight GAP in any index could be 2% or even higher.

(During the past 6 month for example till the beginning of November 2010 we could have 6 big GAP-s against our position, If we opened overnight position at a time, when both the Daily and the Omega prediction pointed in the same direction. The biggest GAP against our position was 2%, the other big GAPs against our position were between 0.9 – 1.6%) During the past half year the market volatility was about average. (GAPs calculated as: (Today’s Open – Previous day’s Close ) / Previous day’s close * 100 [%])

In the coming articles of this series I will write about the relevant technical analysis, and other aspects that can further aid us to decrease the risk and increase the success ratio of our strategy to play overnight position.