> ## Documentation Index
> Fetch the complete documentation index at: https://docs.drivly.dev/llms.txt
> Use this file to discover all available pages before exploring further.
# Overview
**[Drivly Connect](/connect/what-is-connect)** required integrations: Manheim, Kelly Blue Book, CarGurus, Black Book,
NADA Guides, Edmunds. \*At least two book value sources are required for Drivly's Pricing API.
### Drivly Vehicle Pricing:
The Drivly weighted average price calculation takes into account pricing information from a variety of sources, including all major pricing categories, such as MSRP, trade-in, wholesale, private party, retail, and loan values.
```yaml theme={null}
MSRP Average = (nada.msrp + blackBook.msrp) / 2
Trade Average = (cargurus.trade + edmunds.trade.clean + kelley.trade.good +
nada.trade.avgBook + blackBook.trade.totalAvg) / 5
Wholesale Average = (edmunds.wholesale.clean + kelley.wholesale.book +
nada.wholesale.avg + blackBook.wholesale.totalAvg +
manheim.lastSixMonthsPrice) / 5
Private Party Average = (cargurus.privateparty + edmunds.privateparty.clean +
kelley.privateparty.good) / 3
Retail Average = (cargurus.retail + edmunds.retail.clean + kelley.retail.book +
nada.retail.book + blackBook.retail.totalClean) / 5
Loan Average = (nada.loan.book + blackBook.loan.totalValue) / 2
```
```yaml theme={null}
Drivly Price =
(MSRP * w1 + Trade * w2 + Wholesale * w3 + Private Party * w4 + Retail * w5 + Loan * w6)
/ (w1 + w2 + w3 + w4 + w5 + w6)
```
| Source | Trust Score | BBB Rating | User Reviews | Weight |
| ---------------- | ----------- | ---------- | ------------ | ------ |
| CarGurus | 9/10 | A+ | 4.5/5 | 0.23 |
| Kelley Blue Book | 9/10 | A+ | 4.5/5 | 0.23 |
| Edmunds | 8/10 | A+ | 4/5 | 0.18 |
| NADA Guides | 8/10 | A | 4/5 | 0.14 |
| Black Book | 8/10 | A | 3.5/5 | 0.09 |
| Manheim | 8/10 | A | 4/5 | 0.13 |
In this formula, `w1, w2, ... w6` are the weights assigned to the corresponding pricing categories based on the importance, accuracy, and credibility of the sources. These weights are adjusted according overtime to reflect the market.
By calculating these weighted average prices, we normalize the data to find the best approximation of the vehicle's value across various conditions and sources.
```javascript theme={null}
const data = {
vin: 'WP0CD29929S773682',
year: '2009',
make: 'Porsche',
model: '911',
trim: 'Turbo',
msrp: {
nada: 139300,
blackBook: 139300,
},
trade: {
cargurus: 65262,
edmunds: {
outstanding: 70972,
clean: 66809,
average: 59872,
rough: 50853,
},
kelley: {
good: 61891,
baseGood: 61081,
},
nada: {
base: 50525,
baseAvg: 47675,
baseRough: 44300,
book: 66550,
avgBook: 63700,
roughBook: 60325,
},
blackBook: {
baseClean: 62150,
baseAvg: 57125,
baseRough: 48855,
totalClean: 64875,
totalAvg: 60025,
totalRough: 51955,
totalWithHAVClean: 64875,
totalWithHAVAvg: 60025,
totalWithHAVRough: 51955,
},
},
wholesale: {
edmunds: {
outstanding: 59867,
clean: 56356,
average: 50504,
rough: 42897,
},
kelley: {
book: 69319,
base: 68509,
fair: 57788,
good: 64039,
veryGood: 67663,
excellent: 72600,
baseGood: 63229,
},
nada: {
high: 57701,
avg: 55476,
low: 53276,
baseLow: 45550,
baseAverage: 47750,
baseHigh: 49975,
},
blackBook: {
baseClean: 61850,
baseAvg: 56825,
baseRough: 51975,
totalClean: 64575,
totalAvg: 59725,
totalRough: 55075,
totalWithHAVClean: 64575,
totalWithHAVAvg: 59725,
totalWithHAVRough: 55075,
},
manheim: {
lastSixMonthsPrice: 68500,
},
},
privateparty: {
cargurus: 85801,
edmunds: {
outstanding: 81594,
clean: 76612,
average: 68310,
rough: 57516,
},
kelley: {
good: 62871,
baseGood: 62061,
},
},
retail: {
cargurus: 88522,
edmunds: {
outstanding: 87681,
clean: 82120,
average: 72853,
rough: 60805,
},
kelley: {
book: 74852,
base: 74042,
msrp: 142150,
baseFairPurchasePrice: 71909,
totalFairPurchasePrice: 72719,
fairPurchasePriceLow: 67907,
fairPurchasePriceHigh: 77530,
},
nada: {
base: 57250,
book: 73275,
},
blackBook: {
baseClean: 72500,
baseAvg: 63475,
baseRough: 57950,
totalClean: 75225,
totalAvg: 66375,
totalRough: 61050,
totalWithHAVClean: 75225,
totalWithHAVAvg: 66375,
totalWithHAVRough: 61050,
},
},
loan: {
nada: {
base: 45475,
book: 61500,
},
blackBook: {
baseValue: 61825,
totalValue: 64275,
totalWithHAV: 64275,
},
},
}
```
```javascript theme={null}
function calculateWeightedAverages(prices, weights) {
let weightedSum = 0
let totalWeights = 0
prices.forEach((price, index) => {
weightedSum += price * weights[index]
totalWeights += weights[index]
})
return weightedSum / totalWeights
}
const weights = [0.23, 0.18, 0.23, 0.14, 0.09, 0.13]
const msrpAverage = (data.msrp.nada + data.msrp.blackBook) / 2
const tradeAverage =
(data.trade.cargurus +
data.trade.edmunds.clean +
data.trade.kelley.good +
data.trade.nada.avgBook +
data.trade.blackBook.totalAvg) /
5
const wholesaleAverage =
(data.wholesale.edmunds.clean +
data.wholesale.kelley.book +
data.wholesale.nada.avg +
data.wholesale.blackBook.totalAvg +
data.wholesale.manheim.lastSixMonthsPrice) /
5
const privatePartyAverage =
(data.privateparty.cargurus + data.privateparty.edmunds.clean + data.privateparty.kelley.good) / 3
const retailAverage =
(data.retail.cargurus +
data.retail.edmunds.clean +
data.retail.kelley.book +
data.retail.nada.book +
data.retail.blackBook.totalClean) /
5
const loanAverage = (data.loan.nada.book + data.loan.blackBook.totalValue) / 2
const averagePrices = [
msrpAverage,
tradeAverage,
wholesaleAverage,
privatePartyAverage,
retailAverage,
loanAverage,
]
const weightedAveragePrice = calculateWeightedAverages(averagePrices, weights)
console.log('Weighted Average Price:', weightedAveragePrice)
```
***
## Incorporating the standard deviation to understand pricing confidence
For the Drivly Price, we can calculate the average value for each pricing category, and then take a weighted average based on the relevance of each category.
**Drivly Price** is a weighted average of `Trade-in`, `Wholesale`, `Private Party`, and
{' '}
`Retail values`.
```bash theme={null}
(50,525+66,550 + 48,855+64,875 + 65,262 + (50,853+70,972)/2 + (61,081+61,891)/2) / 7
≈ 58,704
```
```bash theme={null}
(45,550+57,701 + 51,975+64,575 + (42,897+59,867)/2 + (57,788+72,600)/2 + 68,500) / 7
≈ 57,889
```
```bash theme={null}
(85,801 + (57,516+81,594)/2 + (62,061+62,871)/2) / 3
≈ $72,566
```
```bash theme={null}
(57,250+73,275 + 57,950+75,225 + 88,522 + (60,805+87,681)/2 + (67,907+77,530)/2) / 7
≈ $72,807
```
```bash theme={null}
(0.4 * 58,704) + (0.2 * 57,889) + (0.2 * 72,566) + (0.2 * 72,807)
≈ $62,954
```
Next, we'll calculate the weighted average considering their relevance. In this example, we assume weights of `40%` for Trade-In, `20%` for Wholesale, `20%` for Private Party, and `20%` for Retail, but these can be adjusted as required based on market data.
Drivly Price = `(0.4 * 58,704) + (0.2 * 57,889) + (0.2 * 72,566) + (0.2 * 72,807) ≈ $62,954`
The Drivly Price, would be approximately `$62,954` for the vehicle with the given VIN `WP0CD29929S773682`.
1. Calculate the mean price: `(58,704 + 57,889 + 72,566 + 72,807) / 4 = $65,241.5`
2. Find the squared differences:
* `(58,704 - 65,241.5)^2 = 42,735,969.25`
* `(57,889 - 65,241.5)^2 = 53,919,863.25`
* `(72,566 - 65,241.5)^2 = 53,534,497.25`
* `(72,807 - 65,241.5)^2 = 57,387,798.25`
3. Average the squared differences: `(42,735,969.25 + 53,919,863.25 + 53,534,497.25 + 57,387,798.25) / 4 = 51,894,532`
4. Take the square root of that average: `√51,894,532 ≈ $7,206`
The standard deviation of the pricing data is approximately **\$7,206**. This measure gives an idea of how widely spread or dispersed the average prices are.
| Category | NADA | Black Book | CarGurus | Edmunds | KBB | Manheim | Drivly |
| ------------- | ------------- | ------------- | -------- | --------------- | ------------- | ------- | -------- |
| MSRP | 139,300 | 139,300 | - | - | - | - | - |
| Trade-In | 50,525-66,550 | 48,855-64,875 | 65,262 | 50,853-70,972\* | 61,081-61,891 | - | - |
| Wholesale | 45,550-57,701 | 51,975-64,575 | - | 42,897-59,867\* | 57,788-72,600 | 68,500 | - |
| Private Party | - | - | 85,801 | 57,516-81,594\* | 62,061-62,871 | - | - |
| Retail | 57,250-73,275 | 57,950-75,225 | 88,522 | 60,805-87,681\* | 67,907-77,530 | - | - |
| Loan | 45,475-61,500 | 61,825-64,275 | - | - | - | - | - |
| STD | - | - | - | - | - | - | \$7,206 |
| Drivly Price | - | - | - | - | - | - | \$62,954 |
* For Edmunds, the range is based on the conditions: Rough, Average, Clean, and Outstanding.
***
### Drivly Price with confidence interval that takes the standard deviation into account.
Using a single standard deviation (which typically accounts for about 68% of the data) to create a confidence interval:
**Drivly Price**: \$62,954 ± \$7,206
This means that we expect approximately 68% of the actual prices to fall within the range of \$55,748 to \$70,160. This gives users a sense of the variability and accuracy of the Drivly Price.
For higher confidence, we can use two standard deviations (about 95% of the data) for the interval. So, approximately 95% of the actual prices should fall within the range of \$48,542 to \$77,366.
This wider interval provides an even higher level of confidence in our pricing accuracy.
***
### Drivly Price with STD and Weighted Price algorithm:
```javascript theme={null}
function calculateWeightedAverages(prices, weights) {
let weightedSum = 0
let totalWeights = 0
prices.forEach((price, index) => {
weightedSum += price * weights[index]
totalWeights += weights[index]
})
return weightedSum / totalWeights
}
function calculateStandardDeviation(array) {
const n = array.length
const mean = array.reduce((sum, val) => sum + val, 0) / n
const squaredDiff = array.map((val) => Math.pow(val - mean, 2))
const variance = squaredDiff.reduce((sum, val) => sum + val, 0) / n
const stdDev = Math.sqrt(variance)
return stdDev
}
const weights = [0.23, 0.18, 0.23, 0.14, 0.09, 0.13]
const weightedAveragePrice = calculateWeightedAverages(averagePrices, weights)
const standardDeviation = calculateStandardDeviation(averagePrices)
const accuratePrice = weightedAveragePrice - standardDeviation
const pricingConfidence = (1 - standardDeviation / weightedAveragePrice) * 100
console.log('Weighted Average Price:', weightedAveragePrice)
console.log('Accurate Price:', accuratePrice)
console.log('Pricing Confidence Score:', pricingConfidence.toFixed(2) + '%')
```
How it works:
1. We first calculate the average price for each pricing type (MSRP, Trade, Wholesale, Private Party, Retail, and Loan) and stores them in variables: `msrpAverage`, `tradeAverage`, `wholesaleAverage`, `privatePartyAverage`, `retailAverage`, and `loanAverage`.
2. These average prices are then combined into an array `averagePrices`.
3. The `calculateWeightedAverages` function computes the Weighted Average Price using the `averagePrices` array and the given weights.
4. Finally, the code prints the Weighted Average Price.
Outputs:
**Weighted Average Price:** \$64,257.955
**Drivly Price:** \$63,984.721 *\*considering the standard deviation*
**Pricing Confidence Score:** 99.57% *\*high confidence level pricing accuracy*
Drivly Price takes the `Weighted Average Price` into account by adjusting it based on the `standard deviation` (variability) of the prices from different sources. The intention of this adjustment is to provide a more reliable and realistic estimation of the car's price, acknowledging the inherent uncertainty and variability.
These values indicate that the initial Weighted Average Price is $64,257.95, the Drivly Price (considering the standard deviation) is calculated to be approximately $63,984.72, and the Pricing Confidence Score is 99.57%, emphasizing the high confidence level in the estimated accurate price.
***
Coming soon!
### Adding in more variables to improve accuracy
Expanding the dataset to include normalized Vehicle History, condition reports, OEM Build Options and high value features, and additional market data (Drivly Listings API).
1. Vehicle History and Condition:
* Add vehicle's mileage, maintenance history, and wear and tear.
* Create a function to calculate condition adjustments based on the provided data from Drivly Connect providers.
* Modify the price estimates by applying the condition adjustments to each pricing type's average.
2. Options and features:
* Include additional pricing properties in the `data` object to account for specific options and features.
* Create a function to calculate adjustments for each option and feature.
* Modify the price estimates by applying the options and features adjustments to each pricing type's average.
3. Market data:
* Utilize Drivly's Listings API and Drivly Connect providers to add regional market data like local demand, supply, and car price trends.
* Create a function to calculate market adjustments based on the provided data.
* Modify the price estimates by applying the market adjustments to each pricing type's average.
Here's an example of how we could implement part of the algorithm for `condition` using a `mileage` adjustment:
```javascript theme={null}
// Update data object with car's condition (mileage) and a sample condition-based adjustment
const data = {
// ...pre-existing data...
condition: {
mileage: 40000,
mileage_adjustment: {
mileage_30000: 0.95,
mileage_40000: 0.9,
mileage_60000: 0.85,
//...
},
},
}
function applyConditionAdjustment(basePrice, mileage, mileage_adjustment) {
if (mileage <= 30000) {
return basePrice * mileage_adjustment.mileage_30000
} else if (mileage > 30000 && mileage <= 40000) {
return basePrice * mileage_adjustment.mileage_40000
} else {
// Assuming mileage > 40000 and mileage <= 60000
return basePrice * mileage_adjustment.mileage_60000
}
}
const conditionAdjustedMSRP = applyConditionAdjustment(
msrpAverage,
data.condition.mileage,
data.condition.mileage_adjustment
)
// ... Do this for other pricing types
// Update the averagePrices array with condition-adjusted prices
// ...
// Proceed with the existing algorithm to calculate the weighted average price and adjusted price
```