On-line portfolio selection using multiplicative updates

On-line portfolio selection using multiplicative updates

ID:37658757

大?。?08.15 KB

頁數(shù):23頁

時(shí)間:2019-05-27

On-line portfolio selection using multiplicative updates_第1頁
On-line portfolio selection using multiplicative updates_第2頁
On-line portfolio selection using multiplicative updates_第3頁
On-line portfolio selection using multiplicative updates_第4頁
On-line portfolio selection using multiplicative updates_第5頁
資源描述:

《On-line portfolio selection using multiplicative updates》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫

1、MathematicalFinance,8(4):325-347,1998.On-LinePortfolioSelectionUsingMultiplicativeUpdatesDavidP.HelmboldRobertE.SchapireComputerandInformationSciencesAT&TLabsUniversityofCalifornia180ParkAvenue,RoomA279SantaCruz,CA95064FlorhamPark,NJ07932dph@cse.ucsc.eduschapire@rese

2、arch.att.comYoramSingerManfredK.WarmuthAT&TLabsComputerandInformationSciences180ParkAvenue,RoomA277UniversityofCaliforniaFlorhamPark,NJ07932SantaCruz,CA95064singer@research.att.commanfred@cse.ucsc.eduAbstractWepresentanon-lineinvestmentalgorithmwhichachievesalmostthe

3、samewealthasthebestconstant-rebalancedportfoliodeterminedinhindsightfromtheactualmarketoutcomes.ThealgorithmemploysamultiplicativeupdaterulederivedusingaframeworkintroducedbyKivinenandWarmuth.Ouralgorithmisverysimpletoimplementandrequiresonlyconstantstorageandcomputi

4、ngtimeperstockineachtradingperiod.WetestedtheperformanceofouralgorithmonrealstockdatafromtheNewYorkStockExchangeaccumulatedduringa22-yearperiod.Onthisdata,ouralgorithmclearlyoutperformsthebestsinglestockaswellasCover'suniversalportfolioselectionalgorithm.Wealsopresen

5、tresultsforthesituationinwhichtheinvestorhasaccesstoadditionalsideinformation."1IntroductionWepresentanon-lineinvestmentalgorithmwhichachievesalmostthesamewealthasthebestconstant-rebalancedportfolioinvestmentstrategy.Thealgorithmemploysamultiplicativeupdaterulederiv

6、edusingaframeworkintroducedbyKivinenandWarmuth[20].Ouralgorithmisverysimpletoimplementanditstimeandstoragerequirementsgrowlinearlyinthenumberofstocks.ExperimentsonrealNewYorkStockExchangedataindicatethatouralgorithmoutperformsCover's[9]universalportfolioalgorithm.The

7、followingsimpleexampledemonstratesthepowerofconstant-rebalancedportfoliostrate-gies.Assumethattwoinvestmentsareavailable.The rstisarisk-free,no-growthinvestmentstockwhosevalueneverchanges.Thesecondinvestmentisahypotheticalhighlyvolatilestock.Onevendays,thevalueofthis

8、stockdoublesandonodddaysitsvalueishalved.Therelativereturnsofthe rststockcanbedescribedbythesequence1;1;1;:::andofthesecondbythe11s

當(dāng)前文檔最多預(yù)覽五頁,下載文檔查看全文

此文檔下載收益歸作者所有

當(dāng)前文檔最多預(yù)覽五頁,下載文檔查看全文
溫馨提示:
1. 部分包含數(shù)學(xué)公式或PPT動(dòng)畫的文件,查看預(yù)覽時(shí)可能會(huì)顯示錯(cuò)亂或異常,文件下載后無此問題,請放心下載。
2. 本文檔由用戶上傳,版權(quán)歸屬用戶,天天文庫負(fù)責(zé)整理代發(fā)布。如果您對本文檔版權(quán)有爭議請及時(shí)聯(lián)系客服。
3. 下載前請仔細(xì)閱讀文檔內(nèi)容,確認(rèn)文檔內(nèi)容符合您的需求后進(jìn)行下載,若出現(xiàn)內(nèi)容與標(biāo)題不符可向本站投訴處理。
4. 下載文檔時(shí)可能由于網(wǎng)絡(luò)波動(dòng)等原因無法下載或下載錯(cuò)誤,付費(fèi)完成后未能成功下載的用戶請聯(lián)系客服處理。