Statistical Learning Models :
Neural Net Daily Trend Identifier : UP [0.81 confidence : uptrend confidence is falling ]
Daily Short-Term Overreaction : DOWN [0.9 confidence]
Filtering Model (4-hour frequency):
Main Trend : UP
Momentum : UP [increasing]
Mean Reversion (4-trading hours frequency) :
4 periods accumulated return (16 hours) : 0.98%
percentile : 70%
9 periods Return (36 hours) : 0.91%
percentile : 58.0%
10 period Volatility (40 hours) : 7.8% annualized
percentile : 22%
Relative Volatility (10 per / 50 per)
percentile : 59%
Markov Regime Switching identifier ( daily close ):
Probability of a strong uptrend regime : 1.4
Probability of a weak uptrend regime : 41.6
Probability of a sideways / weak uptrend regime : 53
Probability of a downtrend / volatile regime : 4
SPY - VIX statistical relationship ( daily close ):
zone : uptrend
percentile :44.0% [ normal ]
Wavelet Discontinuity Watch (daily close ) :
Risk of crash is low but starting to increase.
Higher frequencies starting to increase. Trend change pattern emerging.
Overall commentary for Jan/05:
Neuralnet trend model divergences persist. Model confidence on the uptrend keeps below 0.9. That means trend is weakening/ending.
Main trend still up, with increasing momentum.
Mean reversion setup with 70% probability of win within 16 hours. Rising volatility from low levels (currently at 22% percentile) should make this play even more attractive, albeit less than yesterday, when we reported SPY sitting at its 75% percentile.
MatLab´s Markov model is estimating a higher likelihood of weak uptrend or consolidation. Before a more pronounced correction, downside scenario probability usually shoots up above 20%.
Wavelet model probability of a discontinuity ( for example, a crash) is increasing.
Higher frequencies becoming more dominant on daily data, suggesting higher volatility will be a reality very soon.
Lacking enough Flying-Hours in Quantitative Thinking ? Then try Understanding Daily Reports first.