Version History
Here you can find the changelog of MCarloRisk for Stocks & ETFs since it was posted on our website on 2016-09-28.
The latest version is 19.6 and it was updated on soft112.com on 19 April, 2024.
See below the changes in each version:
version 19.6
posted on 2024-02-04
Feb 4, 2024
Version 19.6
Add link to factor modeling via Vanguard ETFs article, related to this app.
version 19.5
posted on 2024-02-02
Feb 2, 2024
Version 19.5
Add icons on the More view tabs. Fix problem with time series comparison graphs overlaying text for some more advanced cases in the Multiple Regression screen. Allow larger portfolios to be computed without crashing the app.
version 19.4
posted on 2023-10-27
Oct 27, 2023
Version 19.4
Add additional pop tech article links. Increase number of launches between popups.
version 19.3
posted on 2023-10-24
Oct 24, 2023
Version 19.3
Bug fix in prompt popup.
version 19.2
posted on 2023-10-08
Oct 8, 2023
Version 19.2
Add link to new article on backtesting.
version 19.0
posted on 2022-12-05
Dec 5, 2022
Version 19.0
Cleaner caution popup if you don't give share counts to a portfolio of symbols.
version 18.1
posted on 2022-06-20
Jun 20, 2022 Version 18.1
Update app to point to new server.
version 18.0
posted on 2022-05-01
May 1, 2022 Version 18.0
Restore export to PDF feature. Rebuild on newer Xcode.
version 17.8
posted on 2020-08-29
Aug 29, 2020 Version 17.8
Add percent return on yellow graph on the MonteCarlo tab screen for the currently selected price/probability/time. Add yellow background to top floating numbers (yellow graph) since they are getting more involved and difficult to see. When user picks "park" for the floating numbers, also park the upper graph numbers.
version 17.8
posted on 2020-08-29
29 Aug 2020 Version 17.8
Add percent return on yellow graph on the MonteCarlo tab screen for the currently selected price/probability/time. Add yellow background to top floating numbers (yellow graph) since they are getting more involved and difficult to see. When user picks "park" for the floating numbers, also park the upper graph numbers.
version 17.6
posted on 2020-07-19
Jul 19, 2020 Version 17.6
Add 99th percentile curve in light blue to Validate graph and 99th percentile metric to validate residual metrics.
version 17.4
posted on 2020-07-13
Jul 13, 2020 Version 17.4
1. Add diffusion correction option (volatility correction) to Validate screen. Volatility correction is automatically applied if the Validate is run twice with PredCorr feedback loop switch on (it is applied on the 2nd run, "iter = 2" in the user interface). To clear both drift and diffusion correction and start over, press the clear X button in the iter(ation) field. Volatility correction is applied after the drift correction (mean shift) is applied (which is generated on iteration 1 of a PredCorr feedback process). This allows short term bulk backtests to be informed by longer term exhaustive backtests. Volatility correction looks at the residuals from a Validate run and adjusts model volatility after the model is generated to avoid over- or under-estimating risks. A technical paper describing this technique in detail will be available soon. The precise amount of volatility correction in ratio form is noted in the user interface. E.g. if volatility corr is noted as 1.2, it means that the model's volatility will be increased by 20% as the diffusion correction.
1a. In the drift (mean shift) part of the predictor-corrector technique in the above noted Validate menu, change internal formulation to a multiplicative drift correction rather than an additive correction. This prevents the central tendency (50th percentile forecast) price from drifting to be less than zero for forward forecasts, much like if you were to multiply 0.99 by itself many times, the value would never go below zero. However, if you were to continually subtract 0.01 from 0.99, the value would eventually go below zero. Negative prices are not so pretty for a model like this.
2. On newer tall phones (iPhone X and up), provide PDF button on upper left of screen next to the notch to capture any screen of the app as a high resolution scalable "vector" PDF file. Unlike an ordinary screenshot which is saved as a bitmap graphic, this vector file can be scaled up to any required resolution for printing or publication. After saving, a menu is presented to let you decide where to send the PDF file using the standard Apple document handler menu (a variety of options are available, depending what apps on your phone handle PDF files).
version 17.3
posted on 2020-07-07
Jul 7, 2020 Version 17.3
Update to latest Apple tech requirements for iOS13.
version 17.1
posted on 2020-06-21
Jun 21, 2020 Version 17.1
Add interpretation in words of the new Christoffersen metrics in the Validate Report to give users a better idea if exceedence of risk boundaries may be independent over time or not. E.g. if a failure on one day implies higher probability of failure on the next day, or not.
version 17.0
posted on 2020-06-14
Jun 14, 2020 Version 17.0
Basic implementation of Christoffersen’s conditional coverage independence (CCI) test for value-at-risk (VaR) backtesting. Similar to Matlab function cci related to the varbacktest object. Analyzes a backtest for sequential violations (sometimes called"exceedences") of the given percentile (in our case 1% and 5%). Tries to answer the question: "If I get a constraint violation or exceedence one day, how likely am I to get an exceedence on the next day?" Includes intermediate calculation report-out for improved understanding. Computed for 1% and 5% risk levels. Automatically computed during exhaustive Validate procedure in the Validate tab. Results in Report section of the Validate tab.
version 16.9
posted on 2020-06-06
Jun 6, 2020 Version 16.9
Add a predictor-corrector method to allow residuals from long term Validate backtest to correct the most recent forecast envelope. Correction is applied as an adjustment to the drift in the drift-diffusion type of model. Detailed example on how to use this feature is available in the Validate tab by tapping: "Predictor corrector example"
version 16.8
posted on 2020-05-24
May 24, 2020 Version 16.8
In the sampling control toolbar, which is accessed by selecting the "days backwd to sample" field to edit within the Monte Carlo tab of the app, allow a power law decay backwards in time as a sample density option. Allow user to control the power factor. Default power factor is 2.0, quadratic decay backwards in time.
Note that "uniform" sampling is still the overall default: give equal weight to all points within the sampling window.
When sampling is set to Power:
With the power factor set to 0, this is equivalent to uniform sampling (the normal app default).
With the power factor set to 1, this is equivalent to linear decay back in time (the 2nd button setting in the sampling toolbar).
With the power factor set to 0.5 this is equivalent to square root decay back in time. Square root decay takes more
time to decay backwards in time than linear.
You are not limited to just these values; you can use any positive value as the power factor.
With the power factor greater than 1, this puts extra weight on more recent points than does linear, and less weight on the points farther back in time.
As power factor increases greater than 1, more weight is put on the recent points within the resampling window,
meaning, you get more recent points in your monte carlo random walk paths that are generated.
Adjusting sampling probability weights by using this new option may be useful for model calibrating, e.g. getting improved model backtests.
version 16.7
posted on 2020-05-16
May 16, 2020 Version 16.7
Correct graphical end point of red green blue PCA vectors to account for for mean offset of point cloud.
version 16.6
posted on 2020-05-12
May 12, 2020 Version 16.6
Display principal components axes centered at the mean of the returns point cloud instead of at the origin, on Correl(ation) point clould graph.
version 16.5
posted on 2020-05-02
May 2, 2020 Version 16.5
Draw up to the first three principal component axes on the correlation plot (Correl menu) in red=PC0 green=PC1 blue=PC2 colors. Simplifying assumptions: axes are centered at the origin instead of at the mean of the return point cloud, which is usually almost at the origin anyway, and axes are scaled after projecting to the flat plane. Each principal component axis is scaled by the associated singular value, but after projecting to the viewing plane, rather than before. This simplifies the calculations and makes the axes look longer than they would otherwise appear, for easier viewing on a mobile screen. Provide a new switch to show data points and principal axes by themselves to avoid visual clutter if needed; this is the [all | PCA] switch. Show points in grey instead of red when in this isolated PCA display mode to avoid visual conflict with the red axis.
version 16.4
posted on 2020-04-29
Apr 29, 2020 Version 16.4
In Principal Components output text, change wording to "singular value" from "eigenvalue."
version 16.2
posted on 2020-04-06
Apr 6, 2020 Version 16.2
For portfolios, show a color contour plot of pairwise returns distributions smoothed by a Gaussian kernel method. Provide a button to switch this contour off and look at raw data points as before. Data points are lightly overlayed on top of the contour for reference. For example, if your portfolio has 3 assets: SPY, AAPL, TSLA, we compute color contour plots for the pairs [SPY, AAPL], and [SPY, TSLA], and [AAPL, TSLA]. Swipe sideways on the contor plot switch to other pairs. Note that this is a plot of daily returns, not daily prices. This is essentially a smoothed version of a 2D histogram of the paired returns distribution (a 2-dimensional probability distribution). The kernel parameter(s) are not optimized in this iteration, but a reasonable kernel bandwidth is chosen to provide a smoothed visual view of the raw returns point data for typical cases.
version 16.1
posted on 2020-03-07
Mar 7, 2020 Version 16.1
Add small web view on bottom of first page of app which appears at app launch, so we can notify users of various interesting topics without releasing a new app build.
version 16.0
posted on 2020-01-26
Jan 26, 2020 Version 16.0
Allow multiple cores of the CPU to be used when computing Monte Carlo studies, as an option. This is especially useful for long Validate studies (exhaustive backtests). The toggle for this is on the Monte Carlo tab page, upper right side. Default 1CPU will work as before, using only 1 core of the CPU. Tap this to switch it to nCPU to use more than 1 CPU core for calculations.
version 7.7
posted on 2010-12-31
version 17.8
posted on 1970-01-01
29 août 2020 Version 17.8
Add percent return on yellow graph on the MonteCarlo tab screen for the currently selected price/probability/time. Add yellow background to top floating numbers (yellow graph) since they are getting more involved and difficult to see. When user picks "park" for the floating numbers, also park the upper graph numbers.