Published
Edited
Apr 20, 2019
currentSupplyExpectedReturns = forwardChangePredictor(daiChanges.slice(-1)[0])
md`#### use the weighted expected returns in an arbitrage pricing theory type model with other weights expectations`
weightedCurrentSupplyExpectedReturns = currentSupplyExpectedReturns * forwardChangeRsquared
slide`This model use the Arbitrage Pricing Theory with two features that explain ${decimalToPercentStr(totalWeights)} of the price change ${growthLookBack} days later. Currently predicting a ${decimalToPercentStr(totalExpectedReturn)} change`
weightedCollateralExpectedReturns = collateralChartR2 * collateralChartCurrentPrediction
totalExpectedReturn = weightedCurrentSupplyExpectedReturns + weightedCurrentSupplyExpectedReturns
totalWeights = collateralChartR2 + forwardChangeRsquared
decimalToPercentStr = dec => dec.toLocaleString('en-us', {
style: 'percent', maximumSignificantDigits: 2
})
mkrCollateralRegression = createLinearRegression(
mkrCloses,
collateralCloses,
collateralFilter
)
createLinearRegression = (xData, yData, filterFn) => {
let interceptChanges = getBackwardChange(yData).slice(growthLookBack-1)
let forwardReturns = getForwardChange(xData).slice(-interceptChanges.length)
let changesChartData = forwardReturns
.map(
(day, idx) => (
{ returns: day, yChange: interceptChanges[idx]}
)
).filter(filterFn)
let changesChartClean = changesChartData.map(data => [data.yChange, data.returns])
let forwardChangeRegression = simpleStatistics.linearRegression(changesChartClean)
let forwardChangePredictor = simpleStatistics.linearRegressionLine(forwardChangeRegression)
let forwardChangeTrendline = changesChartClean.map(data => ({ y: forwardChangePredictor(data[0]), x: data[0] }))
let forwardChangePredictionRule = vz(forwardChangeTrendline)
.rule()
.x('x')
.y('y');
let forwardPredictionline = forwardChangePredictionRule.copy().line().color('none');
let lastPrediction = forwardChangeTrendline.slice(-1)[0].y
let currentPrediction = forwardChangePredictor(yData.slice(-1)[0])
let chartSupplyRule = vz(changesChartData)
.rule()
.y('returns')
.x('yChange')
.point();
let forwardChangeRsquared = simpleStatistics
.rSquared(
changesChartClean, simpleStatistics.linearRegressionLine(forwardChangeRegression)
)
let chart = vz.layer(chartSupplyRule, forwardPredictionline)
.width(900)
.height(600)
.plot()
return { chart, forwardChangeRsquared, lastPrediction, currentPrediction }
}
Purpose-built for displays of data
Observable is your go-to platform for exploring data and creating expressive data visualizations. Use reactive JavaScript notebooks for prototyping and a collaborative canvas for visual data exploration and dashboard creation.
