Skip to content
#small language model Open access

Bar Types Are Separated by Which of Three Rulers They Hold Constant — For All Five the Declared Ruler Gave the Smallest Coefficient of Variation, Yet the Rank Correlation Is 1.0000 for All Five With a Range of 0.0000, So Correlation Cannot Classify Them [F054]

Sep 2026 · Zenodo (CERN European Organization for Nuclear Research)

Abstract

足を作るとは、何かを一本あたり一定に保つことである。一分足は時間を、ティック足は出来事の数を、レンジ足は変位の絶対量を一定にしている。ではある足種がどの時計なのかは、どうやって決まるのか。名前で決めるのでは、名前を付け替えれば答えが変わってしまう。本稿の分離子は、どの物差しに対して一定にしたか、である。 四銘柄(USDJPY・EURUSD・GBPUSD・XAUUSD)の2019年から2024年までのティックから五つの足種を作った。T 一分足、H 一時間足、K ティック足(N ティックごと)、V 出来高足(出来高の合計が N になるごと)、R レンジ足(経路の長さの合計が S になるごと)である。N は 5,120・4,534・4,780・9,046 ティック、S は 511.6875・340.5750・488.2750・1749.0050 pips で、一時間足と本数がそろうように取ってある。物差しは時間(秒)・ティックの本数・経路の長さ(pips)の三つで、一定さは一本あたり増分の変動係数(標準偏差 ÷ 平均)で測った。基準線として、時刻を実測のまま置いた合成乱数(3 系列)を使った。 五つの足種すべてで、宣言どおりの物差しの変動係数がいちばん小さくなった(5 / 5)。時計で刻む足は時間の 0.0092、ティックで刻む足はティックの 0.0000、経路で刻む足は経路の 0.0011 である。表は対角に並び、次に小さい値との差も、一時間足で 75.31 倍、レンジ足で 212.20 倍とはっきりしている。ティック足と出来高足は三つの物差しすべてで同じ値になったが、これは定義から同じものなので検算であって発見ではない。刻み幅を倍にすると本数はちょうど半分になり、五つとも 0.5000 から 0.5056 である。 ところが順位相関では分類できなかった。五つの足種とも、三つの物差しすべてと 1.0000 の相関を持ち、最大 − 最小は 0.0000 である。理由は単純で、三つの物差しはどれも単調に増えるので、足の番号と必ず 1.0000 で相関するからである。つまり相関は「一緒に増えるか」しか見ておらず、「一本あたり同じだけ増えるか」を見ていない。分類に使えるのは相関ではなく、一定さ(変動係数)である。第3節の表が 0.0000 から 3.3240 まで広がるのに対し、相関の幅は 0.0000 しかない。 そして「時計は誰も動かなくても進む」は、実装では成り立たなかった。一分足の一本の時間の中央は 57.5 秒であって 60 秒ではなく、時間の変動係数も 0.1687 ある。ティックの無い枠に足を作らないので、誰も動かない時間が飛ばされるからである。一時間足は 3599.0 秒・変動係数 0.0092 で、一本が長いほど飛ばされる枠の割合が小さくなる。合成乱数でも最小の変動係数は 0.0000 から 0.1687 で実測と同じであり、分類は市場の性質ではなく足の作り方の性質である。結論として、足種を論じるときに書くべきなのは名前でも相関でもなく、一本あたり増分の変動係数を三つの物差しすべてについて書くことである。そうすれば、その足がどの時計なのかは一行で決まる。 この論文の位置づけ:新しい数学定理も新しい法則も主張しない。時間変形と従属確率過程は古典的な題材である。特定の教材やその著者、販売元は主題ではなく、一切名指していない。売買を論じておらず、成績も損益も測っていない。足種を五つに固定しており、練行足・絶対座標グリッドは入れていない。五つのうち二つは定義から同じもの(ティック足と出来高足)である。刻み幅を一通りに決めており、物差しも三つに固定している。ティックの無い枠に足を作らない規約に依っている。時計で刻む足の比が半分をわずかに超える機序は測っていない。測定は日中・少数銘柄・四銘柄六年に限られ、月次の時系列モメンタムのようにより長い保有で文書化されている結果を否定するものではない。予測を主張しない。 作成にあたって:本稿の着想と内容は、著者自身の考察に基づくものです。文章の構成整理や英訳、集計に用いたスクリプトの作成には AI(大規模言語モデル)の助力を得ました。最終的な内容の解釈や誤りがあれば、それらはすべて著者の責に帰します。お気づきの点があれば、ご教示いただければ幸いです。 ----- Making a bar means holding something constant per bar. A minute bar holds time, a tick bar holds the event count, a range bar holds the absolute move. So how is it settled which clock a given bar type is? Deciding by the name will not do, since renaming it would change the answer. The separator in this paper is which ruler was held constant. From ticks for four instruments (USDJPY, EURUSD, GBPUSD, XAUUSD) from 2019 to 2024, five bar types were built: T, minute bars; H, hourly bars; K, tick bars of N ticks each; V, volume bars closing each time volume sums to N; and R, range bars closing each time the path length sums to S. N is 5,120, 4,534, 4,780 and 9,046 ticks and S is 511.6875, 340.5750, 488.2750 and 1749.0050 pips, set so that the counts match the hourly bars. The three rulers are time in seconds, the tick count, and the path length in pips, and constancy is measured as the coefficient of variation of the increment per bar (standard deviation divided by the mean). A synthetic random walk (3 draws, times left as measured) serves as the baseline. For all five bar types the declared ruler gave the smallest coefficient of variation (5 out of 5). Bars cut by the clock give 0.0092 on time, bars cut by ticks give 0.0000 on ticks, and bars cut on the path give 0.0011 on the path. The table is diagonal, and the gap to the next smallest is clear too: a factor of 75.31 for hourly bars and 212.20 for range bars. Tick bars and volume bars give identical values on all three rulers, but they are the same thing by definition, so that is a check rather than a finding. Double the step and the count halves exactly, running from 0.5000 to 0.5056 for all five. Rank correlation, however, could not classify them. All five bar types correlate at 1.0000 with all three rulers, a range of 0.0000. The reason is simple: all three rulers increase monotonically, so each necessarily correlates at 1.0000 with the bar index. Correlation only asks whether they rise together, not whether they rise by the same amount per bar. What classifies is not correlation but constancy: the table in Section 3 spreads from 0.0000 to 3.3240, against a correlation range of 0.0000. And "a clock advances even when nobody moves" did not hold in the implementation. The median duration of a minute bar is 57.5 seconds, not 60, and the coefficient of variation of its duration is 0.1687, because no bar is created for a slot with no ticks and the time when nobody moves is skipped. Hourly bars give 3599.0 seconds with a coefficient of 0.0092, the longer bar leaving a smaller share of skipped slots. On the random walk the smallest coefficient runs from 0.0000 to 0.1687, as measured, so the classification is a property of how bars are made rather than of the market. The conclusion: what should be written when discussing a bar type is neither its name nor a correlation, but the coefficient of variation of the increment per bar on all three rulers — that settles which clock it is in one line. Scope of this paper: no new mathematical theorem and no new law is claimed; time change and subordinated processes are classical subjects. No real course, author, or vendor is the subject here, and none is named. No trading is discussed, and neither performance nor profit and loss is measured. Five bar types are fixed, with renko and absolute-grid bars not included. Two of the five are identical by definition (tick bars and volume bars). One step size is fixed and three rulers are fixed. It relies on the convention of not creating a bar for a slot with no ticks. Why the clock-cut bars exceed a half slightly is not measured. The measurement is intraday and covers four instruments over six years, and does not contradict results documented at longer holding periods, such as monthly time-series momentum. No prediction is claimed. Acknowledgement: The ideas and content of this paper stem from the author's own considerations. Assistance from an AI (a large language model) was used for structuring, English translation, and writing the scripts used for the measurement. Any remaining errors or misinterpretations are solely the author's. Feedback and corrections are sincerely appreciated.

View source

Similar papers

#small language model Dataset Open access Oct 2026

Socratic guiding questions in synthetic arithmetic data: matched LoRA runs (revision v2)

Supporting data, adapters, predictions and code for the article *Low-Cost LoRA Fine-Tuning of Small Language Models for Multi-Step Arithmetic Reasoning* by Jake O'Grady, Asena Isik Gürhan, Chee Fong Ting and Effirul Ramlan (University of Galway). We generated 20,000 GSM8K-derived arithmetic problems with step-by-step s...

O'Grady, Jake, Gürhan, Asena Isik, Chee, Fong Ting et al. · 465 citations
#computer vision Open access Jun 2016

Software Development in Startup Companies: The Greenfield Startup Model

The results are packaged in the Greenfield Startup Model (GSM), which explains the priority of startups to release the product as quickly as possible, and the need to shorten time-to-market, by speeding up the development through low-precision engineering activities.

Carmine Giardino, Nicolò Paternoster, M. Unterkalmsteiner et al. · 178 citations · ⚡14
#computer vision Open access Oct 2016

Software Startups - A Research Agenda

Software startup companies develop innovative, software-intensive products within limited timeframes and with few resources, searching for sustainable and scalable business models.

M. Unterkalmsteiner, P. Abrahamsson, Xiaofeng Wang et al. · 157 citations · ⚡17
#machine learning Review Open access Oct 2016

“Failures” to be celebrated: an analysis of major pivots of software startups

This study conducts a case survey study based on the secondary data of the major pivots happened in 49 software startups, and demonstrates that customer need pivot is the most common among all pivot types.

Sohaib Shahid Bajwa, Xiaofeng Wang, Anh Nguyen-Duc et al. · 127 citations · ⚡15
#computer vision Review Open access May 2015

A survey study on major technical barriers affecting the decision to adopt cloud services

The comparison of adopter and non-adopter sample reveals three potential adoption inhibitor, security, data privacy, and portability, which underlines the importance of the technical and security perspectives for research investigating the adoption of technology.

Nattakarn Phaphoom, Xiaofeng Wang, S. Samuel et al. · 111 citations · ⚡8
#computer vision Open access Feb 2018

Lean Internal Startups for Software Product Innovation in Large Companies: Enablers and Inhibitors

This study investigates how Lean internal startup facilitates software product innovation in large companies and identifies its enablers and inhibitors, and shows the potential of the method-in-action framework to investigate the Lean startup approach in non-startup context.

Henry Edison, Nina M. Smørsgård, Xiaofeng Wang et al. · 78 citations · ⚡6

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.