A DolphinDB-Based Study of Price Wicking in Cryptocurrency Markets

From theoretical and market microstructure perspectives, price wicking is mainly driven by the following factors:

• Informed Trading: Large institutions or traders with private information buy or sell an asset in large quantities over a given period, causing sharp short-term price fluctuations.
• Insufficient Market Liquidity: Especially in markets with low trading volume, such as certain small cryptocurrency pairs, large orders entering or exiting the market can cause prices to move sharply in an instant.
• Market Manipulation: Some market participants may deliberately influence prices through large-scale buying or selling, thereby triggering price wicking.
• Leverage and Forced Liquidation Mechanisms: High leverage in perpetual futures contracts, combined with forced liquidation rules, means that even a small price deviation can trigger cascading liquidations. Liquidation orders then amplify the market move, creating a "price wicking-liquidation-price wicking" feedback loop.

In addition, an exchange's liquidation mechanism directly affects the magnitude of market impact:

• Full Forced Liquidation: Once triggered, the entire position is immediately liquidated at market price. This mechanism is simple but blunt. When a large position is liquidated, it can drain liquidity instantly and is highly likely to trigger a chain of further liquidations.
• Partial Forced Liquidation: Exchanges such as Binance use this mechanism. The system first attempts to close part of the position so that the margin ratio returns above the maintenance threshold, which in theory reduces market impact. However, during the extreme market conditions of October 2025, prices fell so rapidly that the partial forced liquidation mechanism failed. The system was forced to trigger forced liquidations repeatedly, producing an effect equivalent to full forced liquidation ^[1].

In trading contexts, a "price wick" is essentially a type of extreme OHLC bar pattern: it has a long shadow, a small body, and a price range significantly larger than normal. In terms of its origins, the qualitative description of this type of pattern in classic candlestick analysis can be traced back to Steve Nison's systematic study of candlestick charting techniques ^[2]. Traditional definitions, however, are mostly qualitative. To quantify the definition of a "price wick" intuitively, this article uses a pattern-ratio method and represents a price wick as follows:

• The shadow length is significantly larger than the body length;
• The shadow length accounts for a large proportion of the total OHLC bar range;
• The total OHLC bar range reaches a specified absolute or relative threshold, filtering out ordinary bid-ask spread fluctuations.

The shadow, body, and range are defined as follows:

• Upper shadow length: upper = high - max(open,close)
• Lower shadow length: lower = min(open,close) - low
• Body length: body = |open -close|
• Range: amplitude = high - low

The specific formula is as follows:

$\mathrm{upper(lower)}\ge a\times \mathrm{body}$

$\mathrm{upper(lower)}\ge b\times \mathrm{amplitude}$

$\mathrm{amplitude}\ge c\times \frac{\mathrm{open}+\mathrm{close}}{2}$

VPIN (Volume-Synchronized Probability of Informed Trading) is a high-frequency indicator for measuring order flow toxicity, proposed by Easley et al. in 2011 ^[3]. It measures the degree of information asymmetry in order flow, or "toxicity". The core idea is to "measure time by trading volume": time is divided into volume buckets with equal trading volume, and order imbalance is measured by calculating the difference between aggressive buy and sell orders in each bucket.

The formula is as follows:

$\mathrm{VPIN}=\frac{\sum _{\tau =1}^n|V_{\tau }^{\text{buy}}-V_{\tau }^{\text{sell}}|}{n\times V}$

Here, V is the fixed trading volume of each bucket, and n is the number of buckets in the calculation window. The numerator is the absolute value of the difference between aggressive buy and sell trading volume, representing order flow imbalance.

Before calculating VPIN, you need to determine the number of buckets and the fixed trading volume per bucket. In this article, the bucket volume V is set to the average daily trading volume over the past 20 trading days divided by N, with N set to 50 buckets per day.

$V=\frac{\text{total trading volume over the past 20 trading days}}{20\times N}$

Next, calculate and sum the imbalance between aggressive buy and sell orders in each bucket. Finally, normalize the result by dividing it by the number of buckets n in the calculation window and the fixed trading volume V.

From the perspective of basic indicators, price wicking is often accompanied by changes in the order book depth ratio and the funding rate:

• Order Book Depth Ratio: The ratio of cumulative resting order volume from the best bid to the Nth bid level to cumulative resting order volume from the best ask to the Nth ask level. When this ratio drops sharply, it indicates that downside support has weakened and that there is less resistance to a price decline ^[4].

  Therefore, two related indicators can be constructed:

  Depth Ratio

  $\mathrm{depthRatio}=\frac{\mathrm{bidDepth}}{\mathrm{askDepth}}$

  Normalized Imbalance

  $\mathrm{depthRatioNorm}=\frac{\mathrm{bidDepth}-\mathrm{askDepth}}{\mathrm{bidDepth}+\mathrm{askDepth}}$

  Here, bidDepth and askDepth represent the cumulative order quantity (or notional amount) from bid/ask level 1 through bid/ask level N, respectively.

• Funding Rate: The degree to which the perpetual futures contract price deviates from the spot price. An extremely negative funding rate indicates that the market is highly bearish, but it may also signal a reversal wick caused by a short squeeze. An extremely positive funding rate suggests overcrowded long positions, which can easily trigger a long liquidation cascade.

Based on minute-level OHLC bar data and the criteria described in Chapter 1, the relevant code is as follows:

symList = ["BTCUSDT"]
sySource =  "Binance-Futures"
dataKline = select * from loadTable("dfs://CryptocurrencyKLine","minKLine") where 
      symbol in symList and symbolSource = sySource  order by eventTime 
//If the data uses UTC + 8h, standardize it to UTC time:
// update dataKline set eventTime = temporalAdd(eventTime, -8, 'h')
//Downsample the data:
//barMinutes = 5
//dataKline = select eventTime, sym, sySource, first(open) as open,
//	max(high) as high, min(low) as low, last(close) as close, sum(volume) as volume
//    from dataKline group by bar(eventTime, barMinutes*60*1000) as eventTime
//The shadow is significantly longer than the real body
update dataKline set body = abs(close - open)
update dataKline set amplitude  = (high - low)
update dataKline set upper = (high - max(open,close))
update dataKline set lower = (min(open,close) - low)
update dataKline set avgPrice = (open + close)/2
//Detection criteria; modify as needed
a,b,c = 5.0,0.5,0.5
pinbar = select * from dataKline where lower > body * a and lower > amplitude * b
pinbar_upper = select * from dataKline where upper > body * a and upper > amplitude * b 
pinbar.append!(pinbar_upper)
pinbar = select * from pinbar where amplitude > avgPrice*c*0.01 order by eventTime

Line 6 standardizes the time zone to UTC. Lines 20 and 21 define the core criteria for detecting price wicking. To use the upper shadow instead, use upper. If you need to downsample OHLC bars at different frequencies, modify barMinutes accordingly.

Using 5-minute OHLC bar data for BTCUSDT on Binance as an example, the sample results are as follows:

The results show that OHLC bars matching price wicking are often accompanied by large trading volume, especially compared with the average trading volume for the day. For example, at 2025.10.10T21:20:00.000 in the figure, the low price reached 101,516.5, about 7,000 below the opening price of 108,888, before rebounding to a closing price of 107,747.3. The trading volume in this interval was roughly an order of magnitude higher than in other periods.

The VPIN-related functions and volatility analytic functions are encapsulated in the CalcVPINModule module. For the complete code, see the attached file CalcVPINModule.dos. The module contains the following functions:

Usage:

1. Place the module file in the <path returned by getHomeDir()>/modules path, and then call it with use:

   use CalcVPINModule

2. Set the parameters as needed, and choose either parallel or serial execution to calculate VPIN values across multiple days:

   symList = [`BTCUSDT,`ETHUSDT,`XRPUSDT,`ADAUSDT] 
   targetSource = "Binance-Futures"  // Or "OKX-Futures"
   dataType = "trade"  //"trade" or "minKLine"
   d,N,n = 20,50,10  //Use the previous d days and N buckets per day to calculate V, with a window size of n
   //Run parallel calculations across multiple trading days:
   Date = 2025.10.01..2025.10.30   //UTC time
   result = peach(calcVPINForAllSym{,symList,targetSource,dataType,d, N,n}, Date)
   res = select * from unionAll(result) order by symbol, eventTime
   //If memory is limited, run the calculation serially
   res = table(1000:0, 
           `eventTime`symbolSource`symbol`diffQty`VPIN`price_open`price_high`price_low
           `price_close`price_range`bucket_volume,
           [TIMESTAMP, SYMBOL, SYMBOL, DOUBLE, DOUBLE, DOUBLE, DOUBLE, DOUBLE, DOUBLE, 
           DOUBLE, DOUBLE])
   for(targetDate in Date){
       res_VPIN = calcVPINForAllSym(targetDate,symList,targetSource,dataType,d, N,n)  
       res.append!(res_VPIN)
   }

3. res is the table containing the VPIN calculation results. You can also call the calcVPINVolCor function to obtain basic volatility and correlation results:

   result = calcVPINVolCor(res)
   result

4. Finally, use DolphinDB's built-in plot function to draw the chart:

   sym = "ADAUSDT"
   plotData = select * from res where symbol = sym
   plot(plotData[`volatility_range`VPIN],plotData.eventTime,sym +"：VPIN vs Volatility",extras={multiYAxes: true}) 
   plot(plotData[`volatility_log`VPIN],plotData.eventTime,sym +"：|Log Return| (%)",extras={multiYAxes: true})

Sample Results:

Based on aggregated trade data, the following example shows the VPIN and volatility results for October 2025:

The sample volatility correlation results are as follows:

To show the relationship between VPIN and volatility more clearly, the following comparison charts are plotted:

The charts show a clear synchronization between VPIN and volatility for ADAUSDT and ETHUSDT in October 2025, especially around the extreme market event on October 10, 2025. This indicates that rising order flow toxicity is typically accompanied by amplified short-term price volatility. In normal market conditions, however, the linkage between the two is relatively weak. In addition, across cryptocurrencies, BTC showed a relatively weaker relationship between VPIN and volatility during this period, possibly because its larger market capacity and stronger liquidity absorption make it more resilient. By contrast, a smaller cryptocurrency such as ADA, with weaker liquidity and insufficient order book depth, is more likely to translate order flow imbalance into significant price volatility.

Building on the definition of order book imbalance introduced earlier, this section further analyzes the relationship between this metric and price wicking. The basic calculation procedure is as follows. For the complete code, see the attachment CalcDepthRatioModule.dos.

1. Retrieve order book depth data and calculate the imbalance metric based on the definition.
2. Use the last snapshot in each minute to avoid signal dilution caused by averaging multiple snapshots.
3. Calculate future price returns and amplitudes over different window lengths.
4. Use the quantiles of the order book imbalance metric depthRatioQtyDiff for each symbol as thresholds to identify depth imbalance spikes.
5. Then compare the relationship between spikes and future amplitude by calculating ampRatio = future amplitude with spikes / future amplitude without spikes.

Usage:

The steps above are encapsulated into corresponding functions. To run the analysis, call the main function depthRatioAnalysis:

use CalcDepthRatioModule
symList = [`BTCUSDT,`XRPUSDT,`ETHUSDT,`ADAUSDT]
targetSource = "Binance-Futures"
targetDate = 2025.10.10  //Supports an input time range
futureStep = 5  //Number of future amplitude windows
Data, spikeComparisonForSym, corrForSym = depthRatioAnalysis(symList,targetSource,targetDate,futureStep)

After obtaining the results, use DolphinDB's built-in plot function to chart the data for a single symbol:

//Plot data for a single symbol
sym = "ADAUSDT"
plotData = select * from Data where symbol = sym order by timestamp
update plotData set priceChange = (highPrice - lowPrice) / highPrice
// Compare the current depth metric with the current price change
plot(plotData[`priceChange`depthRatioPrice],plotData.timestamp.minute(),
    sym + " priceChange vs depthRatioPrice", extras={multiYAxes: true})
plot(plotData[`priceChange`depthRatioPriceDiff],plotData.timestamp.minute(),
    sym + " priceChange vs depthRatioPriceDiff", extras={multiYAxes: true})
    
// Compare the depth metric with future returns
plot(plotData[`futureRet5`depthRatioPrice], plotData.timestamp, sym + " futureRet5 vs depthRatioPrice",
    extras={multiYAxes: true})
plot(plotData[`futureRet5`depthRatioPriceDiff], plotData.timestamp, sym + " futureRet5 vs depthRatioPriceDiff", 
    extras={multiYAxes: true})

Example Results:

Using the parameters in the example above, the following table shows sample Data output for the order book imbalance metric on October 10, 2025:

After filtering for imbalance spikes, the relationship with the future 5-minute amplitude is shown below. Here, ampRatio is the ratio of the future 5-minute amplitude during spike periods to the future 5-minute amplitude during non-spike periods:

To illustrate the relationship between the order book imbalance metric and price volatility more clearly, the following comparison charts are plotted:

The relationship table for depth spikes and future amplitude shows that when depthRatioQtyDiff experiences a sharp move at the 90th-percentile level within a single symbol, the future 5-minute amplitudes of BTC, ETH, and XRP increase significantly, with ampRatio values of approximately 2.20, 2.00, and 3.42, respectively. This indicates that, for these symbols, depth spikes are indeed associated with subsequent increases in short-term volatility and can provide a degree of early warning for price wick events. By contrast, the ADA results are unusually unstable, indicating potential issues with the data or threshold settings for this symbol. ADA therefore cannot be included directly in the same conclusion.

The two depthRatioPrice and priceChange charts show no stable linear relationship in which the order book depth imbalance signal consistently precedes future price amplitude. In the later period from 21:10 to 22:50, priceChange fluctuated sharply and displayed price wick peaks, accompanied by severe fluctuations in market liquidity as measured by depthRatioPrice. In other words, this metric is suitable for risk and volatility signal detection as well as price wick identification, and can serve as an early warning signal for price wicking.

Event Background:

On the evening of October 10, 2025, U.S. President Donald Trump announced a 100% tariff on all Chinese imports and new export controls on critical software. This unexpected geopolitical escalation immediately triggered risk-off sentiment across global markets. The S&P 500 fell 2.7%, Bitcoin plunged by more than 14%, Ethereum declined by approximately 12%, and smaller altcoins suffered even larger losses, with many dropping 40% to 70% before partially recovering ^[5]. Within several hours between October 10 and October 11, more than USD 19 billion in leveraged positions were liquidated, making it one of the largest single-day liquidation events in crypto history.

At the microstructure level, the crash occurred on a Friday evening in the United States, when institutional market makers were entering the weekend period and market liquidity was near a low point. Bitcoin had just reached a new all-time high of approximately USD 125,000, and the market was crowded with overleveraged long positions, creating highly fragile conditions.

Analysis Using Real Data:

1. Based on the pinbar table (price wick identification result) from Section 4.1, we filter for minute-level OHLC bars on October 10, 2025 that meet the price wicking criteria. Some examples are shown below:

   pinbar_= select * from pinbar where date(eventTime)=2025.10.10 and symbol = "BTCUSDT"

   

   As shown above, prices fluctuated significantly between 21:00 and 22:00, during which many price wicks occurred.

2. Liquidation data is one of the core datasets for studying price wicking. It reflects the precise price levels where major participants target and flush out highly leveraged positions, the cascading effect of serial liquidations, and the real impact of market liquidity depletion. The code example and results are as follows:

   liqDataForSym = select * from loadTable("dfs://CryptocurrencyDay","liquidation")
   update liqDataForSym set eventTime = temporalAdd(eventTime, -8, "H")  //Normalize to UTC time
   liqDataForSymDay = select sum(quantity),side, date,symbolSource from liqDataForSym 
       group by symbolSource,side, date(eventTime) as date
   res1 = select * from liqDataForSymDay order by date
   liqDataForOneDay = select sum(quantity), side,eventTime,symbolSource from liqDataForSym 
       where date(eventTime) = 2025.10.10 group by symbolSource,side, bar(eventTime,1m) as eventTime
   res2 =select * from liqDataForOneDay order by sum_quantity desc

   

   

   The charts above show liquidation data from different exchanges (OKX-Futures and Binance-Futures) for selected dates and intraday minutes. The liquidation volume on October 10, 2025 was far higher than on other dates, with a large number of leveraged traders facing forced liquidation. Sell-side liquidations greatly exceeded buy-side liquidations, highlighting the extreme volatility in market sentiment on that day. On October 10, 2025, liquidation volumes were extremely large between 21:10 and 22:00, corresponding to the severe price volatility around that period. Using XRPUSDT as an example, the relationship between minute-level liquidation data and price changes is shown below:

   

   As shown above, liquidation activity spiked between 21:00 and 22:00, coinciding with sharp price volatility and highlighting the close relationship between liquidations and price movements.

3. Order book depth is an important metric for measuring market liquidity and the market's capacity to absorb price pressure. Because the order book depth data used here covers only a limited number of levels, we use the difference between the best bid and best ask prices to represent market liquidity. A wider spread typically indicates poorer liquidity and higher price volatility risk. The code example and results are as follows:

   //...
   data = select * from loadTable("dfs://CryptocurrencyTick","depth") 
       where symbol = "BTCUSDT" and symbolSource = sySource and eventTime between startTargetTime:endTargetTime
   update data set eventTime = temporalAdd(eventTime, -8, "H") //Normalize to UTC time
   plotData = select eventTime,symbolSource,symbol,(bidQty[0]-askQty[0]) as diffQty from data
   // Overall difference across multiple bid and ask price levels
   // plotData = select eventTime,symbolSource,symbol,each(sum,bidQty-askQty) as diffQty from data
   plot(plotData[`diffQty],plotData.eventTime.minute(),"bid-ask spread",extras={multiYAxes: true})

   

   As shown above, order book depth experienced several large fluctuations during the day, with the most severe movements occurring between 21:00 and 22:10. This reflected an extreme order flow imbalance, and made price wick events more likely.

4. Trading volume is one of the core factors that drives price volatility and is essential for analyzing price wick events. The data used in this section is aggregated trade data provided by the exchange. It includes only regular market orders and excludes special orders such as liquidation and forced liquidation orders. The following are the single-day trading volume summary table and the chart showing the relationship between intraday trading volume and price for the XRPUSDT trading pair:

   

   

   Based on the market trading volume for "XRPUSDT" on that day, trading volume increased sharply after 20:50, indicating a significant shift in market sentiment. A large influx of orders disrupted the supply-demand balance and directly amplified price fluctuations, demonstrating the impact of substantial trading volume on price volatility.