The model slipped 0.1% this week as the shift out of equities into bonds and commodities continues with a 13% transfer this week. Portfolio 2 with the tighter equity screening sold most of the equity percentage. It outperformed the base model as the McClellan oscillator fell all week and shows that the breadth of the rally is falling as fewer stocks are holding up the indices. The model is selling the utility index VPU on Monday. Portfolio positions bonds and commodities 71% (+13%) equity 22% (-13%) cash 6% (0%) PREDICTION OUTPUT (ALL INSTRUMENTS AVERAGE) (AVG) Trading 45.9% 17.7% 3/17/2020 8/21/2020 since 5/1/20 Model portfolio 12.0%(+1.1% this week) Portfolio 1 11.1%(+0.8%) Portfolio 2 13.1%(+1.4%) SPY 16.9%(+0.9% this week) 60%/40% model 6.9%(+0.7% this week) 40%/60% model 10.7%(+0.7% this week)
More bonds and commodities were added this week as the long bond yields continued to increase. The net result was a small 0.3% gain vs an S&P move of 3.8%. The model has definitely started taking equity profits as it dropped equities to half of their previous levels and is positioning for a correction. Monday morning the model gave up on VMW. Portfolio positions bonds and commodities 69% (+12%) equity 24% (-12%) cash 6% (0%) PREDICTION OUTPUT (ALL INSTRUMENTS AVERAGE) (AVG) Trading 47.6% 19.2% 3/17/2020 8/28/2020 since 5/1/20 Model portfolio 12.3%(+0.3% this week) Portfolio 1 11.6%(+0.5%) Portfolio 2 13.3%(+0.2%) SPY 20.7%(+3.8% this week) 60%/40% model 8.7%(+1.1% this week) 40%/60% model 12.7%(+2.0% this week)
Finally another correction to evaluate performance in a down period. The portfolio was down 0.4% while the S&P lost 2.8%. Pretty good drawdown protection. It would seem the model captures 59% of the up moves and limits the drawdown by 85%. In a flat market that would work great, but in an uptrend it doesn't capture enough of the move. Since 5/1, it handily beat both a 60/40 and 40/60 model, but I lag the S&P by 5.9%, which is almost 32% of the move since May. I broke the the model into 3 models, one for each of the equities, bonds and commodity positions. Then I plan to re-balance the portfolio into a 30/30/30 distribution, filling any open positions with index funds like SPY, DBA and BND, as long as they run above the 20 day MA. The model has already started reducing bond positions in favor of equities. The commodity positions remain stable for now, Portfolio positions equity 31.8% (+7.8%) bonds 35.1% (-17.3%) commodities 22.9% (+0.1%) cash 4.2% (-1.8%) since 5/1/20 Model portfolio 12.0% (-0.4% this week) Portfolio 1 11.2%(-0.4%) Portfolio 2 17.9%(-0.5%) SPY 17.9%(-2.8% this week) 60%/40% model 7.6%(-1.1% this week) 40%/60% model 11.0%(-1.3% this week) PREDICTION OUTPUT original model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 49.5% 21.0% 3/17/2020 9/4/2020 Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Optimization 231.9% 4096.4% 1/3/2000 5/27/2020 (ALL INSTRUMENTS AVERAGE) (AVG) Trading 23.1% 6.3% 5/21/2020 9/4/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Optimization 23.9% 273.0% 11/18/2004 5/27/2020 (ALL INSTRUMENTS AVERAGE) (AVG) Trading 15.4% 4.2% 5/28/2020 9/4/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Optimization 9.4% 101.5% 4/10/2007 5/27/2020 (ALL INSTRUMENTS AVERAGE) (AVG) Trading 4.8% 1.3% 5/28/2020 9/4/2020
September is giving us some correction action and the model was negative but outperformed the S&P by 3.1% this month (-1.2% vs -4.3%) Not a whole lot of activity as I continue to transition from the original model to the 3 models for equities, commodities and bonds. The cash position did increase to 10% and the commodity positions remain underfunded likely due to the dollar strength. The only change for monday is getting back into WMT after its 7 day correction, down 9%. Portfolio positions equity 35.5% (+0.3%) bonds 30.8% (-1.0%) commodities 23..4% (+0.5%) cash 10.3% (+6.1%) since 9/1/20 Model portfolio -1.2% (-0.8% this week) SPY -4.3%(-2.4% this week) 60%/40% model -1.8%(-0.9% this week) 40%/60% model -2.6%(-1.4% this week) PREDICTION OUTPUT original model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 44.0% 19.6% 3/17/2020 9/11/2020 Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 18.6% 5.4% 5/21/2020 9/11/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 11.3% 3.3% 5/28/2020 9/11/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 4.5% 1.3% 5/28/2020 9/11/2020 The McClellan Summation still hasn't found a bottom yet
September continue to test the protective action of the model as I had a -0.4% drawdown vs a -5.3% drop on the S&P. It even managed a small 0.8% gain for the week as it shifted more cash into the commodity positions. The original model still has numerous long equity positions on as I transition to the new equity model. The new model wants to add back MSFT and TSLA and cash in on DG(+2.9%) and KR(+3.5%) on Monday. Portfolio positions equity 35.2% (-0.3%) bonds 29.8% (-1.0%) commodities 32.2% (+8.8%) cash 2.8% (-7.5%) Performance since 9/1/20 Model portfolio -0.4% (+0.8% this week) SPY -5.3%(-1.0% this week) 60%/40% model -2.2%(-0.4% this week) 40%/60% model -3.2%(-1.2% this week) PREDICTION OUTPUT original model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 44.0% 20.4% 3/17/2020 9/18/2020 Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 19.4% 6.0% 5/21/2020 9/18/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 14.5% 4.5% 5/28/2020 9/18/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 4.3% 1.3% 5/28/2020 9/18/2020 The summation index tried to rally but ticked lower. The lower RSI of 15 now has a positive divergence and potential upcoming MACD cross suggests we may be looking for a short term bottom.
The model took a big 2.1% drawdown this week, even as the S&P recovered to trim its loss to -0.5%. The big drawdown was primarily due to the dollar strength negatively affecting the commodity portion of the portfolio. Silver had a -14% correction last week, Gold -4.5%. Cash reserve increased 6% as the model waits for some new openings. It wants to add a REIT OHI and take small profits on DG(+2.0%), MLM(+3.1%), and WBA(+0.7%). Portfolio positions equity 34.4% (-0.8%) bonds 33.4% (+3.6%) commodities 23.4% (-8.8%) cash 8.8% (+6.0%) Performance since 9/1/20 Model portfolio -2.5% (-2.1% this week) SPY -5.8%(-0.5% this week) 60%/40% model -2.5%(-0.3% this week) 40%/60% model -3.6%(-0.4% this week) PREDICTION OUTPUT original model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 40.3% 19.5% 3/17/2020 9/25/2020 Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 15.9% 5.2% 5/21/2020 9/25/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 10.0% 3.3% 5/28/2020 9/25/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 3.2% 1.0% 5/28/2020 9/25/2020 The McClellan still hasn't found a bottom yet
A nice recovery from last weeks drawdown with a 1.1% recovery in a volatile week. Equity positions increased with small decreases in bonds, commodities and cash. I believe I have discovered an error in the prediction indicators that will affect performance as it moves forward. The model uses some moving averages, but the way they are interpreted, I don't think it is monitoring the changes to the averages, rather it uses the trained value of the average. That may be why it doesn't seen to trend follow as well as it should and works more l;ike a mean reversion strategy. As the price or average changes, the model looks at the trained value of the averges and likely sells when the price gets too far above the static average value. I'm looking to change it to the %change in the average price, rather than the value of the average price. A small but subtle difference that becomes worse the father out from the training period we get. Portfolio positions equity 37.1% (+2.7%) bonds 33.1% (-0.3%) commodities 22.8% (-0.6%) cash 7.1% (-1.6%) Performance since 9/1/20 Model portfolio -1.4% (+1.1% this week) SPY -4.4%(+1.4% this week) 60%/40% model -2.0%(+1.6% this week) 40%/60% model -2.8%(+0.8% this week) PREDICTION OUTPUT original model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 41.0% 20.6% 3/17/2020 10/2/2020 Only managing 5 remaining positions BEP, BMY, DUK, HLT and VAW Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 17.7% 6.1% 5/21/2020 10/2/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 6.5% 2.3% 5/28/2020 10/2/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 3.1% 1.1% 5/28/2020 10/2/2020 The McClellan oscillator appears to have put in a bottom this week.
The model was rewritten to use the rate of change of the moving averages instead of the actual MA values and retrained for the last 4 years through July 2. Its a subtle change but should keep the prediction valid at any price levels as time goes on. The allocation percentage were also adjusted with the upcoming Nov-April period which has historically favored equities. It now has a 50/30/20 allocation for equities/bonds/and commodities. I'm actually more bullish on commodities due to the expected inflationary pressures which should drive bonds lower as the yeild come up and commodity prices higher. Equities will need to see a big earnings increase in the next 6 months to offset the high P/E which will eventually start holding back price gains (see big tech in September). INPUT CONTRIBUTION ANALYSIS # Contribution Input 1 15.78 % ROC%(ExpAvg(Time Series,ExpAvg Periods),Periods) 1-15 day 2 23.14 % ROC%(Avg(Time Series,Avg Periods),Periods) 15-30 day 3 27.95 % ROC%(Avg(Time Series,Avg Periods),Periods) 30-80 day 4 31.27 % ROC%(Avg(Time Series,Avg Periods),Periods) 150-200 day 5 24.40 % PriceTarget%(Trading Strategy,Target percent change) 0.1-25% 6 33.08 % TrailPrice%(Trading Strategy,Percent Trailing) 0.1-25% Portfolio positions equity 54.6% (+17.5%) bonds 20.7% (-12.4%) commodities 18.7% (-4.1%) cash 6.0% (-1.1%) Performance since 9/1/20 Model portfolio 1.8%(+3.2% this week) SPY -0.7%(+3.7% this week) 60%/40% model -0.6%(+1.4% this week) 40%/60% model -0.6%(+2.2% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 6.2% 22.7% 7/2/2020 10/9/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 4.8% 17.7% 7/2/2020 10/9/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading -0.3% -1.0% 7/2/2020 10/9/2020 The McClellan oscillator really started racing higher this week, similar to the May-June rally.
Very little change this week as the model underperformed the S&P by -0.7%, actually having a small drawdown while the S&P was up 0.2%. The allocations remained almost unchanged with a larger equity position than in September, Portfolio positions equity 54.6% (+0%) bonds 20.6% (-0.1%) commodities 19.4% (+0.7%) cash 5.5% (-0.5%) Performance since 9/1/20 Model portfolio 1.3%(-0.5% this week) SPY -0.5%(+0.2% this week) 60%/40% model -0.5%(+0.1% this week) 40%/60% model -0.5%(+0.1% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 5.6% 19.3% 7/2/2020 10/16/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 4.7% 16.2% 7/2/2020 10/16/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading -0.1% -0.2% 7/2/2020 10/16/2020 The McClellan oscillator slowed its uptrend this week, but still remains positive.
Managed a small gain on a down week. The dollar seems poised on the brink, awaiting the stimulus bill to drop further and the longer bonds are dropping as the rates go up, likely due to inflation concerns. The model is fully invested at this time as we are about to go into the best 6 months of the year. The only 2 changes for Monday are banking an 8.2% gain in FB and stopping out of CWH with a -11% drop in the last 4 days. Portfolio positions equity 55.9% (+1.3%) bonds 22.7% (+2.1%) commodities 20.1% (+0.7%) cash 1.3% (-4.2%) Performance since 9/1/20 Model portfolio 1.4%(+0.1% this week) SPY -1.0%(-0.5% this week) 60%/40% model -0.9%(-0.4% this week) 40%/60% model -0.9%(-0.3% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 6.0% 19.5% 7/2/2020 0/23/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 4.5% 14.5% 7/2/2020 10/23/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading -0.4% -1.3% 7/2/2020 10/23/2020 The McClellan was flat for the week but appears ready to turn up again
I've actually had the software since 2004. The backtesting and forward testing functions helps give some confidence in the strategy since the neural network never really tells you what the trading rules are, just the weight of the inputs. It may weigh a 5 period moving average by 40%, but you don't know if that means buy if above, sell below, sell if above by 5% etc. Ward systems group Neuroshell trader series. I use the EOD version, but it even has an interday version available
More downward movement as everything was sold, bonds, commodities and stocks. The S&P lost 5.4% and the portfolio managed a 4.0% loss. Pharmaceuticals and Staples are set to be added Monday with the addition of PFE, LLY, and GIS. Portfolio positions equity 52.9% (-3%) bonds 25.8% (+3.1%) commodities 20.0% (-0.1%) cash 1.3% (0.0%) Performance since 9/1/20 Model portfolio -2.6%(-4.0% this week) SPY -6.4%(-5.4% this week) 60%/40% model -3.2%(-2.3% this week) 40%/60% model -4.3%(-3.4% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading $5.90845 3.5% 10.5% 7/2/2020 10/30/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading $0.94727 1.4% 4.3% 7/2/2020 10/30/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading $-0.48714 -0.5% -1.5% 7/2/2020 10/30/2020 The McClellan fell significantly with the sell off.
Everything gained this past week but the S&P 200 point rally seemed a bit narrow with the heavily weighted stocks strong but the rest of the market less strong (6% vs 7.3%). The model captured about 2/3 of the gain, which is typical since it is made for a lower beta to avoid the drawdowns. Portfolio positions equity 53.1% (+0.2%) bonds 26.6% (+0.8%) commodities 19.1% (-0.9%) cash 0.0% (4.7%) Performance since 9/1/20 Model portfolio +2.1%(+4.7% this week) SPY +0.3%(+6.7% this week) 60%/40% model -0.2%(+3.0% this week) 40%/60% model +0.0%(+4.3% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 5.7% 16.3% 7/2/2020 11/6/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 3.9% 11.2% 7/2/2020 11/6/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading -0.1% -0.4% 7/2/2020 11/6/2020 The McClellan reversed but hasn't moved up nearly as strong as the S&P, not even to the 50% of the October high.
Equities with a big rally Monday on the vaccine news, the the S&P gained another 2.3%, commodities up slightly and bonds down again. The model again captured almost 1/2 of the equity gain, and is still outperforming due to the Sept and Oct drawdowns. I deleted the commodity funds COW and DBA. Since August Soybeans are up 35%, Corn 28%, Wheat 20%, but the fund is up 7%, Somethings broken with the Invesco DB Agriculture Fund. I just cant find a good Ag fund to take advantage of rising food prices. Selling PPL, LUV, and JNJ Monday after nice gains. Portfolio positions equity 54.3% (+1.2%) bonds 26.9% (+0.3%) commodities 15.0% (-4.3%) cash 5.0% (+5%) Performance since 9/1/20 Model portfolio +3.4%(+1.3% this week) SPY +0.3%(+2.3% this week) 60%/40% model +0.6%(+0.8% this week) 40%/60% model +1.3%(+1.3% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 8.8% 24.1% 7/2/2020 11/13/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 4.4% 12.0% 7/2/2020 11/13/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading -0.3% -0.7% 7/2/2020 11/13/2020 McClellan continues to rise but a negative divergence is developing at the all time high close Friday
Posted a small gain on a flat to down week after cutting equities and raising commodities and cash this week. Even bonds had a positive week. Mondays moves will trim Platinum, oil, MGM and GIS, all taking profits. Portfolio positions equity 30.4% (-23.9%) bonds 30.4% (+3.5%) commodities 21.4% (+7.4%) cash 16.5% (+11.5%) Performance since 9/1/20 Model portfolio +4.3%(+0.9% this week) SPY 1.8%(-0.8% this week) 60%/40% model -0.1%(+0.6% this week) 40%/60% model +0.7%(-0.3% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 9.3% 24.1% 7/2/2020 11/20/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 2.5% 6.5% 7/2/2020 11/20/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 0.1% 0.2% 7/2/2020 11/20/2020 McClellan started to roll over at the October high.
Nice 2.4% gain this week as the equity position filled as the cash deployed and the commodities were sold as the precious metals fell despite the dollar weakness. Bonds were flat again with an average yield of 2.6%. The Model wants to take profits on COST but Im overrulling due to the special Dec 2 dividend. Portfolio positions equity 54.8% (+24.4%) bonds 29.2% (-1.2%) commodities 13.9% (-7.5%) cash 2.0% (-14.5%) Performance since 9/1/20 Model portfolio +6.7%(+2.4% this week) SPY 4.2%(+2.4% this week) 60%/40% model +1.6%(+1.7% this week) 40%/60% model +2.5%(+1.5% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 11.3% 27.8% 7/2/2020 11/27/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 3.4% 8.3% 7/2/2020 11/27/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 0.0% 0.0% 7/2/2020 11/27/2020 McClellan really not buying into the thanksgiving rally
Another 1.7% S&P rally this week and the model captured 0.9% of the gain. Bond slipped again this week while we had a commodity rally in Gold, Silver, oil and bitcoin. Platinum logged almost a 10% gain. The bond and commodity positions have been steadily shrinking the past few weeks, partly as a result of the equity gains. Bonds, especially the longer durations were down again and continued their slide since the summer. The program is cashing in on LVS Monday with a 2.6% 1 day scalp Portfolio positions equity 57.9% (+3.2%) bonds 28.9% (-0.3%) commodities 11.1% (-2.8%) cash 2.0% (+0.0%) Performance since 9/1/20 Model portfolio +7.8%(+1.1% this week) SPY 5.9%(+1.7% this week) 60%/40% model +2.0%(+0.4% this week) 40%/60% model +3.3%(+0.8% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 12.3% 29.0% 7/2/2020 12/4/2020 Commodity model ALL INSTRUMENTS AVERAGE) (AVG) Trading 11.8% 27.7% 7/2/2020 12/4/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading -0.3% -0.7% 7/2/2020 12/4/2020 S&P McClellan only up marginally but may be trying to rally with the market.
A 1.0% S&P dip this week and the model slipped 0.2%. Bond gained back a little while commodities were mixed with gains in gold, grains, cattle, and bitcoin while silver, platinum and oil slipped as the dollar rallies a little off it new 52 week low. Monday plan to add BBH, GIS, KMB (biotech and consumer staples, a more defensive posture). Portfolio allocations equity 59.6% (+1.7%) bonds 29.9% (+1.0%) commodities 9.1% (-2.0%) cash 1.4% (-0.6%) Performance since 9/1/20 Model portfolio +7.6%(-0.2% this week) SPY +4.9%(-1.0% this week) BND -0.3%(+0.4%) DJP +2.2(+1.2%) 60%/40% model +1.8%(-0.2% this week) 40%/60% model +2.8%(-0.5% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 11.9% 26.8% 7/2/2020 12/11/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 13.4% 30.2% 7/2/2020 12/11/2020 Bond model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 0.0% -0.1% 7/2/2020 12/11/2020 S&P McClellan rolling over but still in Novembers range
S&P added another 0.8% this week and the model added 2.6%, primarily on the commodity 3.4% gain on the dollar drop. Bond slipped slightly 0.1% while commodities gains in metals, oils, grains, and bitcoin (a 43% move in GBTC) were big as everything was working this week. Cash was raised following the gains this week. Portfolio allocations equity 32.3% (-26.9%) bonds 29.2% (-0.7%) commodities 12.4% (+3.3%) cash 26.0% (+26.6%) Performance since 9/1/20 Model portfolio +10.2%(+2.6% this week) SPY +5.7%(+0.8% this week) BND -0.4%(-0.1%) DJP +3.4(+1.2%) 60%/40% model +2.1%(+0.3% this week) 40%/60% model +3.3%(+0.5% this week) PREDICTION OUTPUT Equity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 12.3% 26.5% 7/2/2020 12/18/2020 Commodity model (ALL INSTRUMENTS AVERAGE) (AVG) Trading 21.1% 45.6% 7/2/2020 12/18/2020 Bond model ALL INSTRUMENTS AVERAGE) (AVG) Trading -0.2% -0.4% 7/2/2020 12/18/2020 S&P McClellan Mostly flat for the week but drifting lower
